Volatile organic compounds in exhaled breath as non-invasive biomarkers for Gastrointestinal motility disorders: an exploratory study Authors: Kim W. E. Sweerts, José M. Conchillo, Daniel Keszthelyi, Danielle M. Pachen, Agnieszka Smolinska, Zlatan Mujagić Published: 2025 ABSTRACT Background and aimsGastrointestinal (GI) motility disorders are characterized by abnormalities in the motor functions of the GI tract. The diagnostic evaluation of these disorders frequently involves invasive and time-consuming examinations, for which access may be limited. Volatile organic compounds (VOCs) could serve as non-invasive alternative. Therefore, the aim of this study was to explore the potential of exhaled breath VOCs as biomarkers in patients with GI symptoms and potential GI motility disorders.MethodsIn this exploratory, prospective study, breath samples were obtained from patients undergoing ambulatory motility tests as part of routine clinical care. VOCs in exhaled breath were assessed using thermal desorption chromatography-mass spectrometry (TD-GC-MS). The resulting data were subsequently pre-processed and analyzed using machine learning approaches.ResultsHundred participants were included in the analysis, of whom 67 were women (67%), with a median age of 56.5 years (IQR: 29.8). The diagnostic work-up comprised 55 gastric emptying tests, 55 high-resolution esophageal manometries, 3 antroduodenal manometries, and 1 colon manometry. These examinations resulted in 48 motility disorder diagnoses, while 51 patients showed no evidence of motility abnormalities. Fifteen VOCs were identified as most discriminative markers for the presence or absence of GI dysmotility, with a sensitivity of 75%, and specificity of 60%.ConclusionsVOCs in exhaled breath show promise to distinguish patients with GI motility disorders from those without, in a population of patients with GI symptoms. Future research is warranted to further refine and validate these results in a larger cohort and to explore the diagnostic performance of VOCs in specific subtypes of motility disorders.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-025-04568-6. Background Gastrointestinal (GI) motility disorders are characterized by abnormalities in the motor functions of the GI tract and include a diverse group of conditions. Dysmotility can occur in the entire GI tract including the esophagus, stomach, small intestine, and large intestine and motility disorders are commonly categorized in organ-specific subtypes, such as esophageal dysmotility, gastric dysmotility, enteric dysmotility, and colonic dysmotility [1]. These subtypes are associated with distinct clinical diagnoses, such as achalasia, ineffective esophageal motility, gastroparesis (GP), chronic intestinal pseudo-obstruction, or slow-transit constipation [2]. Symptoms differ between conditions but are often non-specific, making accurate diagnosis and disease management difficult. The diagnostic evaluation of GI motility disorders frequently involves invasive and time-consuming examinations, for which access may be limited. Many conditions require the insertion of a catheter for assessments such as an esophageal manometry, antroduodenal manometry (ADM), and colon manometry performed in expert centers [3]. Also, the evaluation of the tests requires specialized expertise, and patients often experience the examinations as uncomfortable. While GI symptoms are very prevalent in the general population, GI dysmotility disorders are not, and preselecting patients at high risk of having a motility disorder to undergo invasive testing is challenging in daily clinical practice. Consequently, there is a need for non-invasive and more widely available diagnostic techniques in the field of GI motility disorders, specifically to distinguish patients with GI symptoms with and without GI dysmotility. Alterations in GI motility are associated with a range of metabolic changes within the GI tract. These can arise from altered physiological processes, changes in food processing, and disruptions in composition and activity of the gut microbiota. A portion of the metabolic products produced in the GI tract are volatile and can enter the blood stream, eventually being exhaled via the lungs. Volatile organic compounds (VOCs) can be produced endogenously and exogenously. They are produced by different biological processes such as the intestinal or liver metabolism and gut microbiota and can be released outside the body via exhaled breath, but also via urine, blood, saliva and feces [4, 5]. Consequently, metabolic changes linked to GI motility disorders may be detectible through VOCs in exhaled breath. If confirmed, these VOCs could serve as non-invasive biomarkers to aid diagnostic process of GI motility disorders. In literature, there are several studies that have examined the diagnostic potential of VOCs for different GI and hepatology disorders such as colon cancer, liver diseases, inflammatory bowel disease (IBD) and irritable bowel syndrome (IBS) [6–9]. However, whether VOCs may be useful to assess GI motility disorders has not been studied yet. The aim of this study was to explore the potential of exhaled breath VOCs as biomarkers in patients with GI symptoms and potential GI motility disorders. More specifically, we aimed to determine and validate a specific set of VOCs that could distinguish patients with and without GI motility disorders within a population presenting with similar GI symptoms. Methods Study design A prospective exploratory study was conducted, consisting of 100 patients with GI symptoms suspected of motility disorders for which motility testing was indicated between October 2023 and December 2024 at Maastricht University Medical Centre (MUMC+). The ethical committee of Maastricht University Medical center and Maastricht University, The Netherlands, has reviewed and approved this study protocol (METC 2023 − 0103) and all study procedures were executed in compliance with Good Clinical Practice Guidelines and according to the revised Declaration of Helsinki [10]. All patients provided informed consent. Study participants All patients who were ≥ 18 years old, with a suspicion of GI motility disorder, and underwent GI motility assessment in the context of standard clinical care were eligible for inclusion if written informed consent was provided. Only patients who were < 18 years old or did not provide informed consent were excluded for this study. The predominant GI symptom for which patients were referred was retrieved from the patient health records based on medical notes. GI motility testing consisted of the following motility tests: esophageal manometry, gastric emptying test using 13C-octanoic acid breath test, ADM, or colonic manometry. Detailed protocols and reference values for each procedure are provided in the Supplementary Methods. The study population was divided into two groups based on their motility test results, i.e. presence or absence of a motility disorder. Sample collection and analysis Breath samples were obtained during outpatient clinic visits at Maastricht University Medical Centre. All breath samples were collected within 30 minutes after completion of the motility test. Exhaled breath was collected using the ReCIVA® Breath Sampler (Owlstone Medical, Cambridge, UK), which directs the breath through Breath Biopsy® Cartridges packed with a dual sorbent mix of Tenax TA and Carbograph 5TD (¼” × 3½” stainless-steel tubes coated for inertness; Markes International Ltd, Llantrisant, UK). Each cartridge received 1 L of breath at a controlled flow rate of 250 mL per minute. To prevent ambient air contamination, participants inhaled purified air delivered via the CASPER™® Portable Air Supply unit (Owlstone Medical). Participants breathed normally into the system, and no breath-hold was enforced at any stage. Sampling proceeded until the target volume of 1 L of end-tidal air was reached, which was typically achieved within a single uninterrupted exhalation sequence; therefore, repeated attempts were not required. Immediately after sampling, all tubes were dry-purged on the same day to remove residual moisture and stored at 4–8 °C under sealed conditions. The storage duration prior to analysis ranged between 1 and 9 months. Identical procedures were used to obtain environmental control samples from the testing area. In total, 100 breath samples and 32 ambient air and system blanks were acquired. Analysis of both breath and ambient air samples was performed using a Q Exactive™ GC Orbitrap™ Mass Spectrometer (Thermo Fisher Scientific Inc., USA). Before analysis, TD tubes were purged in a TD-100 thermal desorption unit (Markes International, UK) to eliminate moisture and stabilize VOC content. Samples were initially desorbed at 250 °C for 15 min and focused onto a cold trap (U-T12ME-2 S, C4–C32 range, Markes International) held at 5 °C. Trapped VOCs were then rapidly heated to 250 °C for 5 min and transferred via helium to a GC column (Stabilwax®-DA, Thermo Fisher) using a 5:1 split. The GC oven was programmed from 40 °C (1 min hold) to 270 °C at 10 °C/min. The mass spectrometer scanned a mass range of m/z 30–450 with a resolution of 60.000. Electron ionization was conducted at 70 eV, and both ion source and transfer line were maintained at 270 °C. Instrument tuning, calibration, and leak tests were performed prior to each analytical run. Analytical performance was monitored using QC samples, consisting of mixture of 10 VOCs analyzed every seventh run, complemented by two procedural blanks (empty TD tubes and clean glass tubes subjected to full processing). These controls allowed continuous assessment of background signals, carryover and instrument stability. No baseline drift, new peaks or systematic contamination was observed across batches. All samples were run under identical instrument settings with randomized order, ensuring stable acquisition and excluding time-related analytical bias [11, 12]. Data analysis Data preprocessing was performed using MZmine version 4.3, which included mass detection, chromatogram building, deconvolution, alignment, and duplicates check. This workflow ensured robust and reproducible extraction of spectral features from the breath VOC data. The resulting feature intensity matrix was exported for further statistical evaluation. Before statistical analysis, the data was normalized using probabilistic quotient normalization for better data comparison. For exploratory data analysis, an Unsupervised Random Forest (URF) algorithm was applied to evaluate natural clustering patterns in the dataset. Proximity matrices generated from the URF were subjected to Principal Coordinate Analysis (PCoA) to visualize group separation and potential clustering between individuals with motility disorders and those without, further referred to as controls. To identify discriminatory features between groups, a Random Forest (RF) classifier was employed. Cross-validation and bootstrap resampling (using so called out-of-bag, OOB samples) were used to validate model performance and minimize overfitting. More, specifically to evaluate model performance, we used repeated random subsampling cross-validation (Monte-Carlo cross-validation). At each iteration, the full dataset was randomly split into 80% training and 20% validation, preserving class proportions. This procedure was repeated 1000 times, generating 1000 independent train/validation partitions. Within each iteration, the validation subset contained only samples unseen during model training, preventing information leakage. A RF classifier was trained in each training subset, and out-of-bag (OOB) samples provided an additional internal performance estimate. Repeating the split 1000 times ensured stable, sampling-independent estimates of classification performance and reduced sensitivity to any single data partition. The RF classifier was trained using 1000 trees, with 20 variables randomly sampled at each split (mtry = 20). The splitting criterion was set to maximise node purity (Gini index). The most important variables were selected using variable importance scores obtained from RF, calculated on the training set. For each variable, values were permuted across OOB samples, and the resulting increase in misclassification rate was averaged across all trees to generate an importance score. Variables were ranked by these scores, and the top-ranked variables were retained for further interpretation. Classification performance was evaluated in the validation set using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and F1 and further visualized by plotting the receiver operating characteristic (ROC) curve to assess model discriminative ability. The optimal threshold was selected using Youden Index. Compounds identification was performed using the NIST Mass Spectral Search Program in Xcalibur version 4.3 using mass spectral and retention time matching. Results A total of 100 participants were included in the analysis, of whom 67 were women (67%). The median age was 56.5 years (IQR: 29.8). The predominant GI symptoms in the total population were nausea and/or vomiting (n = 31), dysphagia (n = 23), constipation (n = 3), diarrhea (n = 1), excessive fullness (n = 3), (postprandial) abdominal pain (n = 13), retrosternal pain/gastro-esophageal reflux symptoms (n = 24), and belching (n = 2). The predominant symptoms were generally similar between the two groups; however, retrosternal pain/ gastro-esophageal reflux symptoms were more common in the group without a motility disorder compared to those with a motility disorder (Table 1). Table 1Predominant Gastrointestinal symptoms in all patients and by study groupsPredominant symptomTotal population (n = 100)Motility disorder(n = 48)aNo motility disorder(n = 51)aP valueNausea and/or vomiting, n (%)31 (31)15 (31)15 (29)1.000Dysphagia, n (%)23 (23)15 (31)8 (16)0.095Constipation, n (%)3 (3)1 (2)2 (4)1.000Diarrhea, n (%)1 (1)1 (2)0 (0)0.485Excessive fullness, n (%)3 (3)1 (2)2 (4)1.000(Postprandial) abdominal pain, n (%)13 (13)8 (17)5 (10)0.380Retrosternal pain/ gastro-esophageal reflux symptoms, n (%)24 (24)6 (13)18 (35)0.010Belching, n (%)2 (2)1 (2)1 (2)1.000aOne missing motility test Table 1Predominant Gastrointestinal symptoms in all patients and by study groupsPredominant symptomTotal population (n = 100)Motility disorder(n = 48)aNo motility disorder(n = 51)aP valueNausea and/or vomiting, n (%)31 (31)15 (31)15 (29)1.000Dysphagia, n (%)23 (23)15 (31)8 (16)0.095Constipation, n (%)3 (3)1 (2)2 (4)1.000Diarrhea, n (%)1 (1)1 (2)0 (0)0.485Excessive fullness, n (%)3 (3)1 (2)2 (4)1.000(Postprandial) abdominal pain, n (%)13 (13)8 (17)5 (10)0.380Retrosternal pain/ gastro-esophageal reflux symptoms, n (%)24 (24)6 (13)18 (35)0.010Belching, n (%)2 (2)1 (2)1 (2)1.000aOne missing motility test Predominant Gastrointestinal symptoms in all patients and by study groups aOne missing motility test Diagnostic testing included 55 gastric emptying tests, 55 high-resolution esophageal manometries, 3 antroduodenal manometries, and 1 colon manometry. Based on these examinations, 48 participants were diagnosed with a motility disorder (23 delayed gastric emptying, 24 esophageal dysmotility, and 1 antroduodenal dysmotility), while 51 showed no evidence of motility abnormalities. The identified motility disorders included: achalasia type I and II, ineffective esophageal motility, absent esophageal contractility, distal esophageal spasms, delayed gastric emptying, and antral hypomotility [2]. One gastric emptying test failed, resulting in one missing test result, i.e. the diagnosis was unclassified for this patient. From the full feature set obtained after preprocessing and normalization, RF analysis identified the 15 most discriminatory VOCs using a variable importance ranking derived from the training data. From the set of 15 discriminatory VOCs, 11 VOCs were putatively identified as phenol, dimethyl sulfide, 2-hexanone, 1-butanol, methyl mercaptan, 1-octene, α-pinene, α-myrcene, ethylamine, undecane, and α-methylstyrene (Table 2). The remaining four VOCs could not be confidently matched to any compounds in the NIST library, despite high-resolution measurements obtained using Orbitrap-MS. Chemical identification followed Metabolomics Standards Initiative (MSI) recommendations. Compounds exhibiting consistent retention behavior and high spectral similarity to library entries were assigned as putatively annotated metabolites (MSI Level 2). Four recurring but unannotated mass features were retained as Level 4 unknowns, due to lack of sufficient library evidence to support structural assignment. In Table 1, fold change (FC) is shown as calculated as the ratio of group means (mean motility disorders / mean controls). The FC presented represents the original (non-log-transformed) fold change, where values > 1 indicate higher abundance in cases and values < 1 indicate higher abundance in controls. To visualize how well the selected features separate the two groups, a PCoA was performed using the proximity matrix generated by the unsupervised RF model based on OOB cases. As shown in Fig. 1, the resulting PCoA score plot illustrates moderate separation between the control group and motility disorder group, with some overlap. The first two principal coordinates explained 32% and 12% of the variance respectively, suggesting that the selected VOCs capture meaningful biological variation linked to disease status. Yet, this PCoA score plot serves solely as an visualization tool to display sample distribution and was not used as a statistical model or basis for classification. Table 2Overview of the 15 discriminated volatile organic compoundsNr.Putative VOCs nameMSI levelRT [min]Count ionFC (motility/controls)1phenol217,394,0410.252dimethyl sulfide21,962,0182,632-hexanone25,0771,0490,441-butanol26,456,0624,75methyl mercaptan21,548,1430,361-octene25,1555,0370,04a7α-pinene24,4991,0543,38α-myrcene25,793,0541,39ethylamine20,959,0731,910undecane25,371,08515,311α-methylstyrene28,8188,0780,812Unknown422,997,1012,513Unknown419,555,0543,314Unknown423,1147,0231,815Unknown410,8189,0133,6aMSI: Metabolomics Standards Initiative; RT Retention time, FC Fold change, If FC > 1 → Upregulated in cases, If FC < 1 It is downregulated in cases*compounds statistically significant with p-value < 0.05 based on rank test and post FDR correction Table 2Overview of the 15 discriminated volatile organic compoundsNr.Putative VOCs nameMSI levelRT [min]Count ionFC (motility/controls)1phenol217,394,0410.252dimethyl sulfide21,962,0182,632-hexanone25,0771,0490,441-butanol26,456,0624,75methyl mercaptan21,548,1430,361-octene25,1555,0370,04a7α-pinene24,4991,0543,38α-myrcene25,793,0541,39ethylamine20,959,0731,910undecane25,371,08515,311α-methylstyrene28,8188,0780,812Unknown422,997,1012,513Unknown419,555,0543,314Unknown423,1147,0231,815Unknown410,8189,0133,6aMSI: Metabolomics Standards Initiative; RT Retention time, FC Fold change, If FC > 1 → Upregulated in cases, If FC < 1 It is downregulated in cases*compounds statistically significant with p-value < 0.05 based on rank test and post FDR correction Overview of the 15 discriminated volatile organic compounds aMSI: Metabolomics Standards Initiative; RT Retention time, FC Fold change, If FC > 1 → Upregulated in cases, If FC < 1 It is downregulated in cases *compounds statistically significant with p-value < 0.05 based on rank test and post FDR correction Fig. 1Principal Coordinate Analysis (PCoA) score plot based on the proximity matrix obtained from Random Forest applied to the 15 selected VOCs using only OOB cases. Individuals with motility disorders (red) and control subjects (blue) formed clusters, suggesting discriminatory potential of the selected features Principal Coordinate Analysis (PCoA) score plot based on the proximity matrix obtained from Random Forest applied to the 15 selected VOCs using only OOB cases. Individuals with motility disorders (red) and control subjects (blue) formed clusters, suggesting discriminatory potential of the selected features Classification performance of the RF model was evaluated using the validation samples from cross-validation and OOB predictions, which simulate an independent test set. The receiver operating characteristic (ROC) curve (Fig. 2) demonstrates good diagnostic performance, with the model showing comparable sensitivity and specificity across multiple decision thresholds, with an overall area under the curve (AUC) of 0.74 (95% CI 0.631–0.823). The optimal threshold based on Youden index led to the sensitivity of 0.75 (95% CI 0.518–0.907) and specificity of 0.6 (95% CI 0.520–0.706), PPV of 0.621 (95% CI 0.500–0.746), NPV of 0.707 (95% CI 0.560–0.844) and F1 for 0.679 (95% CI 0.573–0.784) for the validation set. The curve’s shape suggests the model maintains a good balance between true positive and false positive rates, confirming the robustness of the discriminatory VOCs selected. Fig. 2ROC curve of the RF classifier based on OOB validation samples. The model shows relatively consistent sensitivity and specificity, demonstrating its potential for classifying individuals with motility disorders based on exhaled breath VOC profiles. Area under the curve (AUC) = 0.7495% CI 0.631–0.823) ROC curve of the RF classifier based on OOB validation samples. The model shows relatively consistent sensitivity and specificity, demonstrating its potential for classifying individuals with motility disorders based on exhaled breath VOC profiles. Area under the curve (AUC) = 0.7495% CI 0.631–0.823) Discussion In this study, exhaled breath samples were analyzed to identify VOCs that could discriminate between the presence or absence of GI motility disorders in patients with GI symptoms. The results demonstrated the potential utility of VOCs for screening GI motility disorders. Of a wide range of VOCs measured in exhaled breath ultimately 15 VOCs were identified as most discriminative for the presence or absence of GI dysmotility. This set of VOCs achieved a sensitivity of 75%, a specificity of 60%, and an AUC of 0.74. The findings provide the first evidence that exhaled breath VOC profiling, combined with machine learning, show promise for non-invasively discriminating patients with and without GI motility disorders. To the best of our knowledge, this is the first study to explore the potential utility of exhaled breath VOCs in the diagnostic evaluation of GI motility disorders. This represents an important first step toward making the diagnostic process less invasive and more accessible, potentially allowing assessments to be conducted closer to patients’ homes rather than exclusively in tertiary referral centers. Current diagnostic procedures are often invasive and time-consuming. For instance, scintigraphy, to measure gastric emptying, involves exposure to radiation, requires considerable time, and necessitates the involvement of trained personnel [13]. As a non-invasive alternative, gastric emptying breath tests such as the 13C-octanoic acid breath test have been developed [14]. However, their use remains limited to a few specialized centers, which increases patient travel burden, and the procedure itself still requires a long duration of approximately 4.5 h. Our study provides the first promising evidence that exhaled breath VOCs may serve as a screening tool. This approach could allow diagnostic screening to be conducted regionally, thereby reducing referrals and the number of patients who need to undergo invasive GI motility testing at tertiary care centers. Although the pathophysiological link between esophageal motility disorders and changed VOC profiles may be less clear than for gastric or intestinal motility disorders, the same mechanisms could still contribute. Impaired esophageal motility and stasis of food can lead to changes in the esophageal microbiota. Additionally, reflux-related exposure could change the pH and oxidative stress which can modify VOC generation. From a clinical perspective, esophageal motility testing is more routinely performed compared to gastric and intestinal motility for dysphagia or reflux symptoms, but is also invasive. If VOC analysis could serve as a non-invasive screening tool for this group, it might enhance the diagnostic process and patient comfort. In this study, a panel of 15 VOCs was identified that could discriminate between patients with and without motility disorders, achieving a sensitivity of 75%, and a specificity of 60%. Previous research has already shown promising results for exhaled breath VOCs as a screening tool in the field of gastroenterology. Amongst others, VOCs in exhaled air distinguished advanced adenomas (AA) and colorectal cancers (CRC) from controls [15], achieving sensitivity and specificity percentages comparable to those observed in the present study. Moreover, several studies have explored VOC-based diagnostics in disorders of gut-brain interaction. A systematic review identified 3 studies evaluating breath exhaled VOCs in IBS [9, 16–18]. Reported areas under the curve ranged from 0.62 to 0.99, with Baranska et al. and Van Malderen et al. reporting a sensitivity of 89.4% and 97%, and specificity of 73.3% and 21%, respectively [9, 18]. Of note, Baranska et al. reported a VOC panel with particularly high sensitivity. Our sensitivity and specificity are modestly lower, potentially owing to our heterogeneous patient cohort, encompassing various motility disorders, and the inclusion of controls with GI symptoms rather than asymptomatic healthy controls. Additionally, our use of the more advanced ReCIVA breath sampler, in contrast to the Tedlar bag method used by many previous studies, may explain the differences observed in VOC analysis and results. However, this relatively new technique offers more robust and reproducible measurements. Based on our promising explorative findings, future research should subdivide patients by specific motility disorder diagnosis with sufficient large sample sizes, alongside using a healthy control cohort. This design would give more insights into the diagnostic sensitivity and specificity of VOC panels. Of the 15 discriminative VOCs explored in this study, 11 could be identified and 7 of these have been previously associated with GI disorders. The inability to identify the remaining four VOCs may indicate the presence of novel or less-characterized metabolites, or VOCs that are underrepresented in current spectral libraries. Further structural elucidation using MS/MS fragmentation or complementary databases may be necessary to confirm their identity. The compounds phenol, dimethyl sulfide, butanol, methyl-mercaptan, and undecane have been linked to upper GI conditions. Phenol, and undecane were associated with upper GI malignancies [19, 20]. In prior studies, oxidative stress is suggested as underlying mechanism; however, in the context of our findings, it is possible that these alterations reflect dysbiotic changes resulting from changed GI transit times associated with motility disorders. In addition, elevated levels of dimethyl sulfide, an isomer of butanol, and methyl-mercaptan were previously associated with Helicobacter pylori (H. pylori) infection, and halitosis, respectively [21–23]. Both H. pylori infection and halitosis were linked to shifts in the composition of gastric and oral microbiota and gastric function [24]. Given the highly dynamic nature of the microbiome, its composition may be influenced not only by factors such as infection or antibiotic use, but also by changes in GI motility resulting in VOCs changes. For example, delayed GI motility can enhance bacterial fermentation of dietary proteins and lipids. This process has been linked to increased generation of sulfur-containing metabolites, including dimethyl sulfide, and aromatic compounds such as phenol [25]. α-pinene, 1-octene, and 2-hexanone were linked to inflammatory bowel diseases (IBD) [26–28]. While 1-octene has been linked to IBD in general, α-pinene was more specific associated with active ulcerative colitis. These compounds may arise as by-products of inflammation or as metabolic end-products of microbiome changes accompanying chronic intestinal disease [29]. A similar mechanism could plausibly apply to motility disorders. Furthermore, 2-hexanone was described in the context of an ileoanal pouch compared to an intact colon in patients with ulcerative colitis, finding that 2-hexanone was only found in the intact colon suggesting that it is a colon specific VOCs which supports usefulness for detection of GI tract disorders. In summary, our findings support the hypothesis that altered transit times in GI motility may lead to measurable dysbiosis, reflected in exhaled VOC profiles. Although dysmotility-related dysbiosis is a plausible explanation for these findings, the mechanisms are likely to be multifactorial. Changes in GI transit may influence nutrient availability, and bacterial fermentation, thereby influencing microbial metabolism. In addition, processes such as oxidative stress, or altered host metabolism could also contribute to changes in VOCs, independent of the microbiome. Future studies combining microbiome analysis, motility assessment, and VOC profiling are needed to clarify these mechanisms and confirm causal relationships. A key strength of the present study is the included population, which was derived from standard clinical care, thereby reflecting a real-world patient population and enhancing the clinical relevance of the findings. Specifically relevant, considering that this is the first study to explore the potential of exhaled breath VOCs to distinguish patients with GI motility disorders from those with predominantly the same symptoms, but without GI dysmotility. However, some limitations should be acknowledged. This was an exploratory study representing a first step in using metabolomics to differentiate between patients with and without motility disorders. We acknowledge that further refinement and external validation of our model is required, ideally involving patient cohorts from multiple centers rather than only data from a single tertiary care center. Optimization of sensitivity and specificity in larger cohorts is essential to achieve acceptable minimal performance thresholds. It should be noted that positive and negative predictive values will also depend on the prevalence of GI motility disorders in the tested population, and may differ in routine clinical practice compared to our referral cohort. Additionally, (predominant) symptoms and potential confounding factors were not systematically recorded. The use of validated symptom questionnaires and structured assessment of potential confounders such as smoking, diet and medication use will be essential in future studies to correlate clinical symptoms with VOCs profiles and to elucidate the impact of confounding factors on exhaled VOC profiles. And while 55 patients underwent esophageal manometry, and another 55 underwent a gastric emptying test, only 3 patients underwent an ADM and just 1 patient had a colon manometry, limiting the generalizability of the findings to all types of GI motility disorders and it may have introduced partial verification bias. Future research should aim for a more balanced distribution of diagnostic modalities, which would allow for improvement in differentiation between various GI motility disorders, and measuring intra-individual reproducibility should be taken into account. In conclusion, this explorative study identified a set of 15 exhaled breath VOCs that show promise in distinguishing patients with GI motility disorders from those without, in a population of patients with GI symptoms, achieving a sensitivity of 75% and a specificity of 60%. These findings highlight the potential of VOC analysis as a non-invasive screening tool for GI motility disorders. Future research is warranted to further refine and validate these results in a larger cohort and to explore the diagnostic performance of VOCs in specific subtypes of motility disorders. Supplementary Information Supplementary Material 1 Supplementary Material 1