Predicting Response to Neuromodulators or Prokinetics in Patients With Suspected Gastroparesis Using Machine Learning: The "BMI, Infectious Prodrome, Delayed GES, and No Diabetes" Model.

INTRODUCTION: Pharmacologic therapies for symptoms of gastroparesis (GP) have limited efficacy, and it is difficult to predict which patients will respond. In this study, we implemented a machine learning model to predict the response to prokinetics and/or neuromodulators in patients with GP-like symptoms. METHODS: Subjects with suspected GP underwent simultaneous gastric emptying scintigraphy (GES) and wireless motility capsule and were followed for 6 months. Subjects were included if they were started on neuromodulators and/or prokinetics. Subjects were considered responders if their GP Cardinal Symptom Index at 6 months decreased by ≥1 from baseline. A machine learning model was trained using lasso regression, ridge regression, or random forest. Five-fold cross-validation was used to train the models, and the area under the receiver operator characteristic curve (AUC-ROC) was calculated using the test set. RESULTS: Of the 150 patients enrolled, 123 patients received either a prokinetic and/or a neuromodulator. Of the 123, 45 were considered responders and 78 were nonresponders. A ridge regression model with the variables, such as body mass index, infectious prodrome, delayed gastric emptying scintigraphy, no diabetes, had the highest AUC-ROC of 0.72. The model performed well for subjects on prokinetics without neuromodulators (AUC-ROC of 0.83) but poorly for those on neuromodulators without prokinetics. A separate model with gastric emptying time, duodenal motility index, no diabetes, and functional dyspepsia performed better (AUC-ROC of 0.75). DISCUSSION: This machine learning model has an acceptable accuracy in predicting those who will respond to neuromodulators and/or prokinetics. If validated, our model provides valuable data in predicting treatment outcomes in patients with GP-like symptoms.

Symptom Profile of Patients With Intestinal Methanogen Overgrowth: A Systematic Review and Meta-analysis.

BACKGROUND & AIMS: Archaea constitute one of the main 3 domains of the tree of life, distinct from eukaryotes and bacteria. Excessive luminal loads of methanogenic archaea (intestinal methanogen overgrowth [IMO]) have been implicated in the pathophysiology of various diseases, including constipation. To elucidate the phenotypical presentation of IMO, we performed a systematic review and meta-analysis of the prevalence and severity of gastrointestinal symptoms in subjects with IMO as compared with subjects without IMO. METHODS: Electronic databases, including OVID MEDLINE and Cochrane Database from inception until September 2023, were systematically searched. Prevalence rates, odds ratios (ORs), standardized mean difference (SMD), and 95% confidence intervals (CIs) of symptoms were calculated. RESULTS: Nineteen studies were included (1293 patients with IMO and 3208 controls). Patients with IMO exhibited various gastrointestinal symptoms, including bloating (78%), constipation (51%), diarrhea (33%), abdominal pain (65%), nausea (30%), and flatulence (56%). Patients with IMO had a significantly higher prevalence of constipation as compared with controls (47% vs 38%; OR, 2.04; 95% CI, 1.48-2.83; P < .0001) along with lower prevalence of diarrhea (37% vs 52%; OR, 0.58; 95% CI, 0.37-0.90; P = .01) and nausea (32% vs 45%; OR, 0.75; 95% CI, 0.60-0.94; P = .01). Patients with IMO had higher severity of constipation (SMD, 0.77; 95% CI, 0.11-1.43; P = .02) and lower severity of diarrhea (SMD, -0.71; 95% CI, -1.39 to -0.03; P = .04). Significant heterogeneity was detected. CONCLUSION: Patients with IMO exhibit a higher rate and severity of constipation along with lower rate and severity of diarrhea. The distinct phenotype of patients with IMO should be incorporated in patient-reported outcome measures and further correlated with mechanistic microbiome studies.

Predicting Response to Neuromodulators or Prokinetics in Patients with Suspected Gastroparesis Using Machine Learning: the "BIDnD" Model.

INTRODUCTION: Pharmacologic therapies for symptoms of gastroparesis have limited efficacy and it is difficult to predict which patients will respond. In this study, we implemented a machine-learning model to predict the response to prokinetics and/or neuromodulators in patients with gastroparesis-like symptoms. METHODS: Subjects with suspected gastroparesis underwent simultaneous gastric emptying scintigraphy (GES) and wireless motility capsule (WMC) and were followed for 6 months. Subjects were included if they were started on neuromodulators and/or prokinetics. Subjects were considered responders if their Gastroparesis Cardinal Symptom Index (GCSI) at 6 months decreased by ≥1 from baseline. A machine-learning model was trained using lasso regression, ridge regression or random forest. Five-fold cross-validation was used to train the models and the area under the receiver operator characteristic curve (AUC-ROC) was calculated using the test set. RESULTS: Of the 150 patients enrolled, 123 patients received either a prokinetic and/or a neuromodulator. Of the 123, 45 were considered responders and 78 were non-responders. A ridge regression model with the variables: BMI, Infectious prodrome, delayed GES, no diabetes (BIDnD), had the highest AUC-ROC of 0.72. The model performed well for subjects on prokinetics without neuromodulators (AUC-ROC of 0.83) but poorly for those on neuromodulators without prokinetics. A separate model with GET, duodenal MI, no diabetes, and functional dyspepsia performed better (AUC-ROC of 0.75). DISCUSSION: This machine learning model has an acceptable accuracy in predicting those who will respond to neuromodulators and/or prokinetics. If validated, our model provides valuable data in predicting treatment outcomes in patients with gastroparesis-like symptoms.

Symptomatic Response to Antibiotics in Patients With Small Intestinal Bacterial Overgrowth: A Systematic Review and Meta-analysis.

Background/Aims: We performed a systematic review and meta-analysis evaluating the symptomatic response rate to antibiotics in patients with small intestinal bacterial overgrowth (SIBO). Similarly, we performed a meta-analysis on the symptomatic response to antibiotics in irritable bowel syndrome (IBS) patients with and without SIBO. Methods: MEDLINE, EMBASE, Web of Science, and Cochrane databases were searched from inception to March 2021. Randomized controlled trials and prospective studies reporting dichotomous outcomes were included. Results: There were 6 studies included in the first meta-analysis comparing the efficacy of antibiotics to placebo or no antibiotic. This included 196 patients, of whom 101 received antibiotics and 95 received placebo or no antibiotics. Significantly more patients improved with antibiotics (relative risk [95% CI] = 2.46 [1.33-4.55], P = 0.004). There were 4 studies included in the analysis comparing symptomatic response rates in IBS patients with or without SIBO with 266 IBS patients, of whom 172 had SIBO and 94 did not. The pooled response rate for symptomatic response was 51.2% in the SIBO group vs 23.4% in the no SIBO group, respectively. Significantly more IBS patients with SIBO responded to antibiotics compared to those without SIBO (relative risk [95% CI] = 2.07 [1.40-3.08], P = 0.0003). Conclusions: Antibiotics appear to be efficacious in treating SIBO, although small sample sizes and poor data quality limit this interpretation. Symptomatic response rates also appear to be higher in IBS patients with SIBO. This may be the first example of precision medicine in IBS as opposed to our current empiric treatment approach. Large-multicenter studies are needed to verify the results.

A randomized controlled trial of a proactive analgesic protocol demonstrates reduced opioid use among hospitalized adults with inflammatory bowel disease.

Most hospitalized patients with inflammatory bowel disease (IBD) experience pain. Despite the known risks associated with opioids in IBD including risk for misuse, overdose, infection, readmission, and even death, opioid use is more prevalent in IBD than any other chronic gastrointestinal condition. Most hospitalized IBD patients receive opioids; however, opioids have not been shown to improve pain during hospitalization. We conducted a randomized controlled trial in hospitalized patients with IBD to evaluate the impact of a proactive opioid-sparing analgesic protocol. Wearable devices measured activity and sleep throughout their hospitalization. Chronic opioid users, post-operative, and pregnant patients were excluded. The primary endpoint was a change in pain scores from admission to discharge. Secondary endpoints included opioid use, functional activity, sleep duration and quality, and length of stay. Of 329 adults with IBD evaluated for eligibility, 33 were enrolled and randomized to the intervention or usual care. Both the intervention and control group demonstrated significant decreases in pain scores from admission to discharge (- 2.6 ± 2.6 vs. - 3.0 ± 3.2). Those randomized to the intervention tended to have lower pain scores than the control group regardless of hospital day (3.02 ± 0.90 vs. 4.29 ± 0.81, p = 0.059), used significantly fewer opioids (daily MME 11.8 ± 15.3 vs. 30.9 ± 42.2, p = 0.027), and had a significantly higher step count by Day 4 (2330 ± 1709 vs. 1050 ± 1214; p = 0.014). There were no differences in sleep duration, sleep quality, readmission, or length-of-stay between the two groups. A proactive analgesic protocol does not result in worsening pain but does significantly reduce opioid-use in hospitalized IBD patients.Clinical trial registration number: NCT03798405 (Registered 10/01/2019).

Safety and Tolerability of High-Resolution Esophageal Manometry in Children and Adults.

INTRODUCTION: While high-resolution manometry (HRM) is widely accepted as a safe procedure, no study has assessed the safety profile of HRM in clinical practice. This study aimed to determine the safety and tolerability of HRM and to investigate potential determinants of intolerability. METHODS: We obtained HRM procedure reports, demographics, and clinical data (2005-2022) at a tertiary center using electronic chart review. Our primary outcome was HRM tolerability. Multivariable regression was performed to identify associations between the outcome and covariates including age, sex, race, and comorbidities. RESULTS: A total of 5,107 patients (60.3% female) were included. Of them, 5,050 patients (98.9%) tolerated HRM well and 57 patients (1.1%) did not. Age had a statistically significant effect on tolerance: those younger than 18 years had more than a 5-fold increase in not tolerating HRM compared with those aged 18-79 years (5.77% vs 0.99%; odds ratio [OR] = 5.44, 95% confidence interval [CI] 1.60-18.45; P = 0.007), and those aged 80 years or older were also more likely to terminate HRM (2.43% vs 0.99%; OR = 2.56, 95% CI 1.13-5.76; P = 0.024). While prior foregut surgery had a significant effect on tolerance (OR = 8.06, 95% CI 2.29-28.39; P = 0.001), other factors of race, sex, body mass index, and psychological or cognitive disorders had no significant impact. No serious complications were identified. DISCUSSION: HRM is safe and well-tolerated with approximately 1 in every 100 patients being unable to tolerate HRM. Intolerance was more commonly seen in children and seniors due to minor symptoms of discomfort without serious complications. These data points are crucial to counsel patients in whom HRM is being considered.