[A diagnostic model establishment of adult gastroesophageal reflux disease based on high-resolution manometry parameters of the esophagus].

Abstract

Objective: To establish a diagnostic model for adult gastroesophageal reflux disease (GERD) based on the high-resolution manometry (HRM) parameters of the esophagus. Methods: The clinical data of patients who underwent HRM and 24-hour esophageal pH+impedance examination due to suspected GERD at Jiangsu Province Hospital from January 2021 to October 2024 were retrospectively collected. According to the diagnostic criteria and examination results of GERD, the patients were divided into the GERD group [acid exposure time percentage (AET)>4.2% or total reflux times>80 times] and the non-GERD group, and the HRM parameters of the two groups were compared. Patients were randomly divided into the training set and the validation set in a ratio of 7∶3 using R 4.4. The Youden index maximization method was used to determine the optimal diagnostic cut-off value of a single HRM parameter for diagnosing GERD. The multivariate logistic regression model was used to analyze and screen the influencing factors for diagnosing GERD, and the nomogram of the GERD diagnostic model was drawn. The diagnostic ability and accuracy of the model were evaluated respectively by the area under the receiver operating characteristic curve (AUC) and the calibration curve. Finally, the clinical applicability of the model was determined by the decision curve analysis (DCA). Results: A total of 326 patients were included, among which 77 were in the GERD group, including 48 males and 29 females, with an age of [M (Q₁, Q₃)] 57 (42, 64) years. There were 249 cases in the non-GERD group, including 90 males and 159 females, with an age of 53 (42, 59) years. The age, proportion of males, body mass index (BMI), proportion of cases classified by gastroesophageal junction (EGJ), proportion of cases with ineffective esophageal motility (IEM), and proportion of ineffective swallowing times in the GERD group were all higher than those in the non-GERD group. The gastroesophageal junction contraction index (EGJ-CI), the resting pressure of the lower esophageal sphincter (LESP), and the distal contraction score (DCI) were all lower than those in the non-GERD group (all P<0.05).The HHRM related parameters for diagnosing GERD were EGJ-CI, LESP, DCI, the proportion of ineffective swallowing times and failed peristalsis times. The corresponding optimal cut-off values (sensitivity and specificity) were 23 mmHg·cm (1 mmHg=0.133 kPa) (48%, 86%), 13.4 mmHg (81%, 59%), 1 130 mmHg·s·cm (66%, 60%), 0.15 (53%, 66%), 0.35 (24%, 89%), respectively. The results of the multivariate logistic regression model analysis showed that gender (OR=3.82, 95%CI: 1.69-8.61), BMI (OR=1.28, 95%CI: 1.12-1.46), and EGJ-CI (OR=0.95, 95%CI: 0.92-0.97), EGJ classification type Ⅲ EGJ (OR=6.66, 95%CI: 1.51-29.40), and IEM (OR=6.69, 95%CI: 1.27-35.27) were the influencing factors for the diagnosis of GERD. Model 1 was established by referring to the "Milan Score". The AUC, sensitivity, and specificity of the training set for diagnosing GERD were 0.78 (95%CI: 0.71-0.85), 56%, and 92%, respectively. The AUC, sensitivity, and specificity of the validation set for diagnosing GERD were 0.77 (95%CI: 0.66-0.89), 61%, 82%, respectively; The calibration curves of the training set and the validation set indicate that the model had good calibration ability. The DCA curves of the training set and the validation set suggest that the diagnostic model had good clinical applicability. Model 2 was established based on the data of the Chinese population and the above parameters. The AUC, sensitivity, and specificity of the training set for diagnosing GERD were 0.88 (95%CI: 0.83-0.92), 78%, and 82%, respectively. The AUC, sensitivity, and specificity of the validation set for diagnosing GERD were 0.87 (95%CI: 0.76-0.97), 89%, 80%, respectively. The calibration curves of the training set and the validation set indicate that the model had good calibration ability. The DCA curves of the training set and the validation set suggest that the diagnostic model had good clinical applicability. Conclusions: Gender, BMI, EGJ-CI, EGJ morphological classification, LESP, IEM, DCI and the proportion of failed peristalsis times are the influencing factors for diagnosing GERD. The nomogram model incorporating the above factors can diagnose GERD more intuitively.