Abstract
Objective
A definitive diagnosis of gastroesophageal reflux disease (GERD) depends on endoscopic and/or pH‐study criteria. However, high resolution manometry (HRM) can identify factors predicting GERD, such as ineffective esophageal motility (IEM), esophago‐gastric junction contractile integral (EGJ‐CI), evaluating esophagogastric junction (EGJ) type and straight leg raise (SLR) maneuver response. We aimed to build and externally validate a manometric score (Milan Score) to stratify the risk and severity of the disease in patients undergoing HRM for suspected GERD.
Methods
A population of 295 consecutive patients undergoing HRM and pH‐study for persistent typical or atypical GERD symptoms was prospectively enrolled to build a model and a nomogram that provides a risk score for AET > 6%. Collected HRM data included IEM, EGJ‐CI, EGJ type and SLR. A supplemental cohort of patients undergoing HRM and pH‐study was also prospectively enrolled in 13 high‐volume esophageal function laboratories across the world in order to validate the model. Discrimination and calibration were used to assess model's accuracy. Gastroesophageal reflux disease was defined as acid exposure time >6%.
Results
Out of the analyzed variables, SLR response and EGJ subtype 3 had the highest impact on the score (odd ratio 18.20 and 3.87, respectively). The external validation cohort consisted of 233 patients. In the validation model, the corrected Harrel c‐index was 0.90. The model‐fitting optimism adjusted calibration slope was 0.93 and the integrated calibration index was 0.07, indicating good calibration.
Conclusions
A novel HRM score for GERD diagnosis has been created and validated. The MS might be a useful screening tool to stratify the risk and the severity of GERD, allowing a more comprehensive pathophysiologic assessment of the anti‐reflux barrier.
Trial registration
ClinicalTrials.gov (Identifier: NCT05851482).
Keywords: esophagogastric junction, gastroesophageal reflux disease (GERD), high resolution manometry, Milan score, nomogram
KEY SUMMARY.
Summarize the established knowledge on this subject
Gastro‐esophageal reflux disease (GERD) diagnosis follows endoscopic or pH criteria.
High resolution manometry (HRM) is often performed before pH‐study.
Several anatomical and physiological factors related to GERD can be precisely assessed by HRM.
What are the significant and/or new findings of this study?
In this prospective case‐control study, a novel manometric score, the Milan Score (MS), has been built and externally validated.
The MS might be a useful screening tool to predict pathologic GERD, to stratify disease severity and to achieve a more comprehensive pathophysiologic assessment of the anti‐reflux barrier.
INTRODUCTION
According to the Lyon 2.0 consensus, the diagnosis of gastroesophageal reflux disease (GERD) is based on either endoscopy or ambulatory reflux monitoring criteria, 1 with only an ancillary role for high resolution manometry (HRM) in the diagnostic algorithm. The primary indications for HRM in GERD presentations are for localization of the lower esophageal sphincter (LES) for accurate placement of reflux monitoring catheters, and for exclusion of achalasia and other disorders that mimic GERD. 2 However, since many patients with esophageal symptoms undergo HRM, and catheter‐based or wireless reflux monitoring are often poorly tolerated, expensive and not always available, risk stratification of GERD probability using HRM could be a useful adjunct to esophageal symptom evaluation.
Anatomical and physiological factors favoring gastroesophageal reflux can be precisely and accurately assessed by HRM. A recent technical review of Chicago Classification 4.0 (CC v4.0) evaluating esophagogastric junction (EGJ) metrics on HRM underlined the importance of LES‐crural diaphragm (CD) separation and EGJ‐contractile integral (EGJ‐CI) as parameters to identify impaired EGJ barrier function. 3 Both parameters are demonstrated to be associated with pathologic acid exposure time (AET) on ambulatory reflux monitoring. 4 , 5 Another CC v4.0 technical review summarized the state‐of‐the art about ineffective esophageal motility (IEM), 6 both of which are associated with GERD. 7 , 8 To complement established metrics, the straight leg raise (SLR) maneuver is a dynamic test that can effectively assess impaired EGJ barrier function. A recent multicenter study demonstrated the utility of the SLR maneuver in predicting GERD. The intra‐abdominal pressure (IAP) increase during SLR acts as a “stress test” for the EGJ. A defective EGJ allows retrograde propagation of IAP in the thoracic compartment, thus creating the so‐called “common cavity” phenomenon. 9 , 10 An intra‐esophageal pressure increase of 11 mmHg during SLR is associated with pathologic AET (>6%) in patients with persistent esophageal symptoms suspicious for GERD (sensitivity 79%, specificity 85%). 11
Since HRM parameters that associate with GERD encompass a broad spectrum of GERD phenotypes, we hypothesized that the creation of a model including EGJ morphology, EGJ‐CI value, presence of IEM, and SLR maneuver response could be a useful tool to stratify the risk and severity of GERD in patients undergoing HRM. The aim of this study was to develop a predictive model based on previously collected data, 11 to create a nomogram for a rapid computation of a risk score for GERD probability, the Milan Score (MS), and to test the score in a validation cohort.
METHODS
In this multicenter international prospective observational study, consecutive patients were enrolled between December 2022 and February 2023 in 13 high‐volume Institutions across Europe, North America, and Asia. Inclusion criteria were age between 18 and 75 years and HRM and wireless or catheter‐based reflux monitoring study performed for persistent typical or atypical GERD symptoms off proton‐pump inhibitor (PPI) within 2 weeks of each other. Exclusion criteria consisted of body mass index (BMI) > 35 kg/m2, prior esophageal surgery, eosinophilic esophagitis, scleroderma and achalasia spectrum disorders. According to the Lyon 2.0, an AET >6% on MII‐pH or at least 2 days with AET >6% on wireless pH‐study defined pathologic GERD. 1 The study protocol followed the Declaration of Helsinki principles including informed consent from each participating patient, was approved by the proposing center's ethical committee (identifier: HSR 94/int2021) and sent to the collaborating centers. The study was registered at ClinicalTrials.gov (Identifier: NCT05851482). 12 Collaborating centers obtained local institutional board approval, and data sharing agreements were completed with each center. All authors had access to the study data and reviewed and approved the final manuscript.
Clinical evaluation
Demographic and clinical characteristics, including age, BMI, time of symptom onset, symptoms, PPI use and response, smoking habit, barium swallow and endoscopy findings, were abstracted from each patient's medical records. Symptom burden was assessed using validated self‐report questionnaires completed at or prior to the HRM study, and consisted of GERDQ, 13 GERD Health Related Quality of Life (GERD‐HRQL), 14 and Reflux Symptom Index. 15
High‐resolution esophageal manometry
High‐resolution manometry was performed using a solid‐state catheter with 36 circumferentially incorporated sensors spaced at 1‐cm intervals, and following the Chicago Classification 4.0 (CCv4.0) standard protocol, 16 using each institution's preferred HRM system. HRM was performed by experienced physicians or nurses after an overnight fast. Ten swallows of 5 mL room temperature water were performed in the primary position (upright or recumbent) followed by five swallows in the secondary position. Intact, weak, or failed swallows were defined based on distal contractile integral (DCI) values (≥450, 100–450 and ≤100 mmHg•cm•s, respectively). Patients with >70% weak swallows or ≥50% failed swallows were categorized as IEM. Assessment of LES and EGJ characteristics was performed using LES basal pressure, EGJ‐CI, total and intra‐abdominal LES length, and EGJ morphology. The relationship between LES and CD defined EGJ morphology into 3 subtypes: subtype 1: superimposed LES and CD, subtype 2: LES‐CD separation <3 cm, and subtype 3: LES‐CD separation ≥3 cm. EGJ‐CI was calculated during the reference period using the DCI software tool over the EGJ, and corrected for respiration. 5 Multiple rapid swallows (MRS) consisted of five swallows of 2 mL of water administered at <3 s intervals, and contraction reserve was present when the ratio between MRS DCI and mean single swallow DCI was >1.
The SLR maneuver was performed after the standard HRM protocol as previously described. 11 If elevation of one leg was insufficient to increase the IAP, the patient was asked to raise both legs (double leg raise). In case the patient was not able to elevate both legs, the operator was allowed to help the patient, and asked him to maintain leg elevation for 5 s. The SLR maneuver was considered effective if the IAP increased by 50% during SLR. Intra‐esophageal peak pressure 5 cm above the LES was analyzed both during baseline and SLR maneuver, and an increase of 11 mmHg was considered abnormal.
Esophageal pH and pH‐impedance study
All pH studies were conducted after at least 14 days of PPI withdrawal. Wireless 48 or 96‐h pH‐study or catheter‐based 24‐h pH‐impedance studies were performed according to the institution's preference. The wireless pH‐study capsule (BRAVOTM) was positioned 6 cm above the squamocolumnar junction under endoscopic guidance. Multichannel intraluminal impedance‐pH (MII‐pH) studies were conducted using catheters with eight impedances and one or two pH electrodes. The catheter was calibrated using buffer solutions at pH 4.0 and 7.0 and then inserted trans‐nasally with the pH electrode positioned 5 cm above the proximal margin of the LES. Acidic food and drinks were not allowed during the recording. Duration of meal, symptoms and time spent in the recumbent position were noted either on a diary or on the recorder.
Total, upright and recumbent AET, number of acidic, weakly acidic, and weakly alkaline reflux episodes, DeMeester score, bolus exposure, symptom index (SI), and symptom association probability were recorded. 17 Mean nocturnal basal impedance (MNBI) and post‐reflux swallow‐induced peristaltic wave index were also collected from pH‐impedance studies. 18 , 19
Data collection and statistical analysis
The Research Electronic Data Capture (RedCap) platform was used to upload and store de‐identified data. RedCap is a secure online software designed to collect and manage data from different Institutions.
Numerical variables are reported as median and interquartile range (IQR), while categorical variables are reported as count and percentage. The normality of continuous variables was assessed using the Shapiro–Wilk test. Continuous values were compared between the groups using a nonparametric Kruskal–Wallis test, 20 while Chi‐squared test or Fisher test were used as appropriate for categorical values. The outcome variable was pathologic reflux defined as AET > 6% on reflux monitoring studies, and univariate and multivariable logistic models were performed to evaluate the risk of prediction of pathological reflux.
Based on the cohort of patients enrolled from July 2021 to 20 March22, 11 a multivariable logistic model was built with the four variables that resulted statistically significant in the univariate analysis (IEM, SLR, EGJ‐CI and EGJ morphology) to estimate the odd ratio (OR) and 95% confidence intervals (CI) for each variable in the model. A nomogram including these variables was developed. We performed a bootstrap validation (200 bootstrap resamples) for the statistical model. In order to estimate the accuracy of the model, a second cohort of patients was enrolled to provide external validation. Model accuracy was measured using calibration and discrimination. We assessed calibration by comparing the expected number of abnormal AETs (E) with the observed number (O) for subgroups defined by predictors. The calibration plot was drawn to examine the agreement across deciles of predicted risk in test and validation populations. Discrimination was evaluated by calculating the area under the receiver‐operating characteristic (ROC) curve (AUC) using Harrell's C‐statistic (a C‐statistic of 1 indicates perfect discrimination, whereas 0.5 no discrimination). We interpreted values higher than 0.8 as excellent discrimination, 0.6–0.79 as moderate, and 0.5–0.59 as poor. The calibration of logistic models was also assessed by the Hosmer–Lemeshow test. 21 , 22 A non‐significant p value of the Hosmer–Lemeshow test indicates a good fit.
Additionally, the relationship between MS, AET and MNBI was analyzed using the Pearson correlation coefficient. The statistical model was tested on three different subgroups of patients (typical vs. atypical symptoms, BMI > 30 kg/m2, and PPI responders vs. non responders).
A two‐tailed p‐value <0.05 was considered significant for all statistical tests. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina) and R software version 4.3.0 (The R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
The predictive model was built using 295 patients (median age 50.0 years, median BMI 24.8 kg/m2) in the calibration cohort (Table 1). There were 115 patients (38.9%) with pathologic AET, and 136 (46.6%) had a hiatal hernia on HRM. A detailed description of the calibration cohort has been previously published. 11 Of the four variables analyzed, esophageal peak pressure increase during SLR [OR 18.20 (CI 9.52–34.81) p < 0.0001] and EGJ subtype 3 [OR 3.87 (1.36–11.03) p = 0.01] were significant predictors of AET >6%, while EGJ subtype 2 [OR 2.03 (1.01–4.08) p = 0.05] was not significant but numerically different (Table 2). IEM and EGJ‐CI were not demonstrated to be significant predictors [OR: 2.19 (0.92–5.18) p = 0.07 and OR: 0.92 (0.82–1.03) p = 0.17]. When the model was tested on different subgroups of patients, we found the same significant variables as the general population, with slight variations in the odds ratios, indicating that these characteristics apply to patients with esophageal symptoms and different characteristics.
TABLE 1.
Demographic, clinical, HRM and pH characteristics of the overall patient population according to acid exposure time (AET).
| Calibration cohort | Validation cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Total (n = 295) | AET <6% (n = 180) | AET >6% (n = 115) | p‐value | Total (n = 233) | AET <6% (n = 136) | AET >6% (n = 97) | p‐value | |
| Male, n (%) | 134 (51.0) | 73 (45.6) | 61 (59.2) | 0.031 | 116 (49.8) | 60 (42.1) | 56 (57.7) | 0.047 |
| Age, (years) | 49.0 [14.1] | 48.8 [15.0] | 49.3 [12.9] | 0.922 | 50.92 [14.51] | 48.48 [14.22] | 54.34 [14.30] | 0.003 |
| BMI, (kg/m2) | 26.1 [5.6] | 25.7 [6.3] | 26.6 [4.4] | 0.003 | 22.92 [8.64] | 21.61 [8.67] | 24.62 [8.34] | <0.001 |
| Symptoms duration,(months) | 52.2 [54.3] | 51.4 [57.1] | 53.5 [50.1] | 0.126 | 58.57 [70.77] | 46.99 [59.50] | 75.54 [82.07] | 0.001 |
| Primary typical symptoms, n (%) | 262 (88.8) | 156 (86.7) | 106 (92.2) | 0.143 | 218 (93.6) | 125 (91.9) | 93 (95.9) | 0.284 |
| PPI use, n (%) | 214 (80.2) | 119 (73.5) | 95 (90.5) | 0.001 | 168 (75.7) | 93 (71.0) | 75 (82.4) | 0.057 |
| Endoscopic findings | ||||||||
| Hiatal hernia, n (%) | 150 (60.0) | 72 (48.3) | 78 (77.2) | <0.001 | 101 (44.5) | 42 (31.8) | 59 (62.1) | <0.001 |
| Esophagitis, n (%) | 74 (30.0) | 32 (21.8) | 42 (42.0) | 0.001 | 75 (32.9) | 32 (24.2) | 43 (44.8) | 0.002 |
| Barrett's esophagus | 8 (3.1) | 2 (1.3) | 6 (5.9) | 0.064 | 19 (8.6) | 2 (2.4) | 16 (16.8) | <0.001 |
| HRM and pH characteristics | ||||||||
| HRM system, ManoView®, n (%) | 278 (94.2) | 172 (95.5) | 106 (92.2) | 0.224 | 215 (92.2) | 127 (93.4) | 88 (90.7) | 0.453 |
| EGJ type | <0.001 | <0.001 | ||||||
| 1, n (%) | 158 (53.9) | 118 (66.3) | 40 (34.8) | 115 (49.4) | 86 (63.2) | 29 (29.9) | ||
| 2, n (%) | 94 (32.1) | 50 (28.1) | 44 (38.3) | 78 (33.5) | 45 (33.1) | 33 (34.0) | ||
| 3, n (%) | 41 (14.0) | 10 (5.6) | 31 (27.0) | 40 (17.2) | 5 (3.7) | 35 (36.1) | ||
| Hiatal hernia, n (%) | 136 (46.6) | 61 (34.5) | 75 (65.2) | <0.001 | 143 (62.7) | 69 (52.3) | 74 (77.1) | <0.001 |
| Hiatal hernia size, (cm) | 1.7 [1.4] | 1.3 [1.4] | 1.9 [1.8] | 0.005 | 1.7 [1.3] | 1.0 [1.4] | 2 [1.3] | <0.001 |
| LES total length, (cm) | 2.1 [0.8] | 2.1 [0.8] | 1.9 [0.8] | 0.007 | 2.20 [0.68] | 2.32 [0.74] | 2.02 [0.54] | 0.003 |
| LES intrabdominal length, (cm) | 0.2 [1.0] | 0.5 [1.2] | 0.0 [0.5] | <0.001 | 0.54 [0.75] | 0.70 [0.79] | 0.33 [0.64] | <0.001 |
| EGJ‐CI, (mmHg × cm) | 35.2 [42.5] | 46.0 [41.8] | 22.6 [26.5] | <0.001 | 44.47 [31.59] | 54.01 [34.96] | 31.08 [19.56] | <0.001 |
| Patients with IEM, n (%) | 58 (19.7) | 27.0 (15.1) | 31 (27.0) | 0.010 | 44 (19.4) | 20 (14.9) | 24 (25.8) | 0.041 |
| Acid exposure time, (%) | 4.1 [8.7] | 1.5 [3.0] | 11.3 [11.6] | <0.001 | 7.69 [10.97] | 2.06 [1.71] | 15.58 [13.38] | <0.001 |
| DeMeester score | 16.2 [32.8] | 6.5 [12.0] | 45.7 [40.3] | <0.001 | 32.66 [43.09] | 9.28 [8.67] | 63.76 [50.38] | <0.001 |
| Total reflux episodes. n (%) | 42.0 (37.5) | 34.0 (34.0) | 52.5 (49.3) | <0.001 | 27.00 [21.20] | 19.09 [14.52] | 38.44 [24.01] | <0.001 |
| MNBI | 1947.0 [1787.8] | 2749.0 [1915.6] | 1265.5 [1141.3] | <0.001 | 2294.15 [1339.26] | 3043.01 [1196.34] | 1287.88 [718.90] | <0.001 |
| PSPW index, (%) | 44.0 [36.0] | 65.0 [27.0] | 29.5 [16.7] | <0.001 | 56.03 [25.85] | 67.19 [23.09] | 40.97 [21.48] | <0.001 |
Note: Continuous values are expressed as median [IQR].
Abbreviations: AET, acid exposure time; BMI, body mass index; EGJ, esophago‐gastric junction; EGJ‐CI, esophago‐gastric junction contractile integral; HRM, high‐resolution manometry; IEM, ineffective esophageal motility; LES, lower esophageal sphincter; MNBI, mean nocturnal baseline impedance; PPI, proton‐pump inhibitors; PSPW, post‐reflux swallow‐induced peristaltic wave.
TABLE 2.
Multivariate logistic model analysis for outcome acid exposure time >6%.
| OR (CI 95%) | p‐value | |
|---|---|---|
| IEM | 2.19 (0.92–5.18) | 0.07 |
| SLR maneuver | 18.20 (9.52–34.81) | <0.0001 |
| EGJ‐CI (units = 10) | 0.92 (0.82–1.03) | 0.17 |
| EGJ type | 0.02 | |
| 2 | 2.03 (1.01–4.08) | 0.05 |
| 3 | 3.87 (1.36–11.03) | 0.01 |
Abbreviations: CI, confidence interval; EGJ, esophago‐gastric junction; EGJ‐CI, esophago‐gastric junction contractile integral; IEM, ineffective esophageal motility; OR, odds ratio; SLR, straight leg raise.
A nomogram was created based on the predictive model, as shown in Figure 1. To use the nomogram in clinical practice, the score of each parameter is extracted on the top row titled “points” and summed up to generate the MS. The maximum possible score for each parameter is calculated in the model depending on the corresponding odds ratio. The predictive value is determined by the projection of the MS on the bottom row entitled “predicted value”. Figure 2 demonstrates two examples of use of the nomogram. A MS of 137 points has a percent risk of GERD of 50% (sensitivity 80.4%, specificity 85.2%). Lower scores have higher sensitivity with lower specificity and lower risk of GERD, while higher scores have higher specificity with lower sensitivity, and higher risk of GERD. In our population, only 13 (7.6%) of 171 patients with a MS below 60 had an AET > 6%, while 22 (88%) of 25 patients with a MS above 210 had pathological GERD. In order to simplify the use of the MS, risk rates have been categorized in Table 3.
FIGURE 1.
Nomogram to predict GERD (AET > 6%). EGJ‐CI, esophago‐gastric junction contractile integral; IEM, Ineffective esophageal motility; SLR, straight leg raise.
FIGURE 2.
Practical applications of the nomogram. Figure on the left shows a patient with IEM (27 points), positive SLR (100 points), EGJ‐CI of 15 mmHg × cm (61 points) and EGJ type 1 (0 points). Total Milan score188 and predicted value 85%. AET was 8.5%. Figure on the right shows a patient with normal motility (0 points), negative SLR (0 points), EGJ‐CI of 47 mmHg × cm (52 points) and EGJ type 3 (46 points). Total Milan 98 and predicted value 23%. AET was 2.1%.
TABLE 3.
Risk rate quartiles with corresponding Milan score points and associated risk categories.
| Risk rate | Milan score points | Risk category |
|---|---|---|
| RR < 5% | MS < 36 | Extremely unlikely |
| 5% ≤ RR < 25% | 36 ≤ MS < 100 | Very unlikely |
| 25% ≤ RR < 50% | 100 ≤ MS < 137 | Unlikely |
| 50% ≤ RR < 75% | 137 ≤ MS < 175 | Likely |
| 75% ≤ RR < 95% | 175 ≤ MS < 238 | Very likely |
| RR ≥ 95% | MS ≥ 238 | Extremely likely |
When the linear relationship between MS and AET was inquired, a significant and strong correlation was found (r = 0.601, p < 0.001) (Figure S1), while only a moderate correlation was found with MNBI (r = 0.440, p < 0.001). Also, the MS and AET significantly correlated with the GERD‐HRQL questionnaire (r = 0.158, p = 0.002 and r = 0.184, p < 0.001, respectively).
Of 246 patients submitted for external validation, 13 were excluded due to an ineffective SLR maneuver. The final cohort of patients included 233 patients, 97 (41.6%) of whom had confirmed GERD based on pathologic AET. Patients in the AET > 6% group were older (54.3 vs. 48.5 years, p = 0.003), had a higher BMI (24.6 vs. 21.6 kg/m2, p < 0.001) and had a longer symptom duration (75.5 vs. 47.0 months, p = 0.001). All HRM variables associated with GERD (EGJ type, presence and size of hiatal hernia, LES total and intra‐abdominal length, EGJ‐CI and IEM diagnosis) were significantly different between patients with and without pathologic AET (Table 1).
Predictive cohort
For the testing model, the corrected Harrel c‐index was 0.87. The calibration and the ROC curve plot show the predictive accuracy of the model (Figure 3a). The model‐fitting optimism adjusted calibration slope was 0.94 and integrated calibration index was 0.009. The AUC of the testing model was 0.88 (CI 95% 0.84–0.92) and the Hosmer and Lemeshow test was not statistically significant (p = 0.96). These results were confirmed by the bootstrap validation (c‐index 0.72, R 2 = 0.50) (Figure S2).
FIGURE 3.
(a) ROC model (AUC = 0.88 [CI 95% 0.84–0.92]) and Calibration curve on test population (Hosmer and Lemeshow test p = 0.96), c‐index corrected = 0.870. (b) ROC model (AUC = 0.88 [CI 95%: 0.82–0.93]) and Calibration curve on validation population (Hosmer and Lemeshow test p = 0.07), c‐index corrected = 0.900.
Validation cohort
In the validation model, the corrected Harrel c‐index was 0.90. The calibration plot of the model validation is shown in Figure 3b. The model‐fitting optimism adjusted calibration slope was 0.93 and the integrated calibration index was 0.07, indicating good calibration. The AUC of the validation model was 0.88 (CI 95% 0.82–0.93) and the Hosmer and Lemeshow test was also not statistically significant (p = 0.07). The calibration graph shows a good fit across the entire predicted risk range. The calibration curve lying under the ideal line suggests a moderate underestimation of the validated model.
DISCUSSION
In this multicenter study assessing the ability of HRM to predict pathologic AET and severity of GERD, we validate and demonstrate the reliability of our new proposal of the MS that correlates with pathological AET.
GERD is a complex multifactorial disease with a variety of clinical phenotypes and management options that make its diagnosis and treatment challenging to practitioners and patients alike. 23 According to the Lyon 2.0 consensus, 1 a definitive diagnosis of GERD requires either grade B‐C‐D esophagitis on upper endoscopy or AET >6% on a reflux monitoring study performed off‐PPI. However, ambulatory reflux monitoring, particularly when catheter‐based, may be poorly tolerated, and it limits patients' daily life and activities, potentially impacting the results. Although HRM may be difficult to tolerate, identifying the specific disease phenotypes and the underlying pathophysiological mechanisms may stratify patients with different grades of disease, and therefore help personalize GERD management when confirmed, including surgery for the most severe.
A host of anatomical and pathophysiological risk factors for GERD can be found in most patients with GERD symptoms on HRM, and a manometric study is performed in most patients with suspected GERD, especially when endoscopy is normal. Among these HRM factors, a weak EGJ barrier, hiatal hernia and IEM are well‐known abnormalities that correlate with pathological AET. 3 , 4 , 5 , 6 We built a model that included all potential HRM predictors of GERD based on a previous cohort of patients studied with HRM for suspected reflux. 11 Other HRM metrics such as contractile reserve or different degrees of IEM resulted non statistically significant at the univariate analysis, and were therefore not included in the model. Among the included variables, a positive SLR maneuver and EGJ subtype III had the highest impact (OR 18.20 and 3.87 respectively). The non‐predictive role of the EGJ‐CI was potentially due to its evaluation as a continuous variable; nevertheless, this statistical choice was more adherent with current literature and clinical practice, since a validated cut‐off for EGJ‐CI is still to be found and the competency of the EGJ is better described by a stepwise increase of the EGJ‐CI. 3 IEM was also found to be not significant in the multivariable model, but it was included in the score due to its established role in the pathophysiology of GERD. 6 , 7 , 8
A nomogram was created from the model, with the potential for the creation of a web‐based tool for simple and practical clinical use by practitioners (www.milanscore.com).
When the predictive model was applied to the external validation cohort, it showed good fit (AUC = 0.88, CI 95%: 0.82–0.93), thereby validating the model itself.
The MS could be a useful and simple tool in the hands of the HRM operator to stratify the risk of GERD in order to make a more comprehensive assessment of the reflux barrier prior to escalation of medical therapy or surgery. In our population, 92.4% of patients with a MS below 60 (risk of GERD 10%) had a negative reflux study, while 88% of patients with a MS above 210 (risk of GERD 90%) had a positive reflux study. Moreover, the strong correlation between MS and AET suggests that this novel HRM metric is also able to predict the severity of GERD, although further outcome studies will be necessary to add supplementary evidence to this finding.
The major strengths of this study are the prospective, multicenter design and the inclusion of a large sample of patients with persistent GERD symptoms assessed with validated measures, who underwent a thorough pathophysiological evaluation that included ambulatory reflux monitoring and HRM with provocative maneuvers. SLR was effective in 94.7% of the patients, compared with 81% of the development cohort, 11 indicating that the maneuver can be effectively performed in the majority of patients, alone or with the aid of the operator. The double leg raise variation proved to be feasible when attempted, increasing the value of SLR metrics as important components of the MS.
Our study also has a few limitations. Data on treatment and therapeutic outcomes are missing. Further studies are needed to assess the ability of the MS to predict medical or surgical results. Two variables in the model did not significantly correlate with pathologic GERD defined by AET >6%. Although this is expected with IEM given the uncertain clinical implications of this diagnosis, the analysis of EGJ‐CI as a continuous variable might explain its statistically non‐significant relationship to pathologic GERD. In our study, BMI >35 kg/m2 was an exclusion criterion; therefore, the effectiveness of the score on an obese population has yet to be proven. Also, waist circumference data were lacking in our study. Differences in terms of modality of reflux monitoring (catheter‐based vs. wireless pH) were not explored due to a low number of patients studied with wireless pH (33 patients). Also, the different HRM systems used may have a little impact on the final score.
Finally, since all the procedures were performed by expert operators in high volume centers, the results might not be generalizable with the same accuracy in lower volume centers.
Further studies are needed to determine how the MS performs in different populations (i.e. obese patients, isolated laryngo‐pharyngeal reflux, post‐antireflux surgery) and to validate its ability to predict response to medical or surgical therapy with outcome studies.
CONCLUSIONS
A novel HRM score for GERD diagnosis has been validated. We anticipate the MS to be a useful screening tool to predict pathologic GERD, to stratify disease severity, and eventually make the diagnostic pathway more efficient and GERD treatment more precise and personalized.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
Supporting information
Supporting Information S1
Figure S1
Figure S2
ACKNOWLEDGEMENT
Open access funding provided by BIBLIOSAN.
Siboni S, Sozzi M, Kristo I, Boveri S, Rogers BD, De Bortoli N, et al. The Milan score: a novel manometric tool for a more efficient diagnosis of gastro‐esophageal reflux disease. United European Gastroenterol J. 2024;12(5):552–61. 10.1002/ueg2.12565
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
- 1. Gyawali CP, Yadlapati R, Fass R, Katzka D, Pandolfino J, Savarino E, et al. Updates to the modern diagnosis of GERD: Lyon consensus 2.0. Gut. 2023;21(2):361–371. Epub ahead of print. PMID: 37734911. 10.1136/gutjnl-2023-330616 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Gyawali CP, Roman S, Bredenoord AJ, Fox M, Keller J, Pandolfino JE, et al. International GERD consensus working group. Classification of esophageal motor findings in gastro‐esophageal reflux disease: conclusions from an international consensus group. Neuro Gastroenterol Motil. 2017;29(12). PMID: 28544357. 10.1111/nmo.13104 [DOI] [PubMed] [Google Scholar]
- 3. Kahrilas PJ, Mittal RK, Bor S, Kohn GP, Lenglinger J, Mittal SK, et al. Chicago classification update (v4.0): technical review of high‐resolution manometry metrics for EGJ barrier function. Neuro Gastroenterol Motil. 2021;33(10):e14113. Epub 2021 Mar 2. PMID: 33655610; PMCID: PMC8410874. 10.1111/nmo.14113 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Tolone S, de Cassan C, de Bortoli N, Roman S, Galeazzi F, Salvador R, et al. Esophagogastric junction morphology is associated with a positive impedance‐pH monitoring in patients with GERD. Neuro Gastroenterol Motil. 2015;27(8):1175–1182. 10.1111/nmo.12606 [DOI] [PubMed] [Google Scholar]
- 5. Nicodème F, Pipa‐Muniz M, Khanna K, Kahrilas PJ, Pandolfino JE. Quantifying esophagogastric junction contractility with a novel HRM topographic metric, the EGJ‐Contractile Integral: normative values and preliminary evaluation in PPI non‐responders. Neuro Gastroenterol Motil. 2014;26(3):353–360. Epub 2013 Dec 3. PMID: 24460814; PMCID: PMC4605557. 10.1111/nmo.12267 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Gyawali CP, Zerbib F, Bhatia S, Cisternas D, Coss‐Adame E, Lazarescu A, et al. Chicago classification update (V4.0): technical review on diagnostic criteria for ineffective esophageal motility and absent contractility. Neuro Gastroenterol Motil. 2021;33(8):e14134. Epub 2021 Mar 26. PMID: 33768698. 10.1111/nmo.14134 [DOI] [PubMed] [Google Scholar]
- 7. Triadafilopoulos G, Tandon A, Shetler KP, Clarke J. Clinical and pH study characteristics in reflux patients with and without ineffective oesophageal motility (IEM). BMJ Open Gastroenterol. 2016;3(1):e000126. 10.1136/bmjgast-2016-000126 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Gyawali CP, Sifrim D, Carlson DA, Hawn M, Katzka DA, Pandolfino JE, et al. Ineffective esophageal motility: concepts, future directions, and conclusions from the Stanford 2018 symposium. Neuro Gastroenterol Motil. 2019;31(9):e13584. Epub 2019 Apr 11. PMID: 30974032; PMCID: PMC9380027. 10.1111/nmo.13584 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ribolsi M, Holloway RH, Emerenziani S, Balestrieri P, Cicala M. Impedance‐high resolution manometry analysis of patients with nonerosive reflux disease. Clin Gastroenterol Hepatol. 2014;12(1):52–57. Epub 2013 Jul 23. PMID: 23891920. 10.1016/j.cgh.2013.06.034 [DOI] [PubMed] [Google Scholar]
- 10. Butterfield DG, Struthers JE, Jr , Showalter JP. A test of gastroesophageal sphincter competence. The common cavity test. Am J Dig Dis. 1972;17(5):415–421. PMID: 5024985. [DOI] [PubMed] [Google Scholar]
- 11. Siboni S, Kristo I, Rogers BD, De Bortoli N, Hobson A, Louie B, et al. Improving the diagnostic yield of high‐resolution esophageal manometry for GERD: the "straight leg‐raise" international study. Clin Gastroenterol Hepatol. 2022;19(7):1761.e1–1770.e1. Epub ahead of print. PMID: 36270615. 10.1016/j.cgh.2022.10.008 [DOI] [PubMed] [Google Scholar]
- 12. ClinicalTri.als.gov [Internet] . Validation of HRM score for the GERD diagnosis (HRM nomogram). Identifier NCT05851482. Bethesda: National Library of Medicine (US). [Google Scholar]
- 13. Jones R, Junghard O, Dent J, Vakil N, Halling K, Wernersson B, et al. Development of the GerdQ, a tool for the diagnosis and management of gastro‐oesophageal reflux disease in primary care. Aliment Pharmacol Ther. 2009;30(10):1030–1038. 10.1111/j.1365-2036.2009.04142.x [DOI] [PubMed] [Google Scholar]
- 14. Velanovich V. The development of the GERD‐HRQL symptom severity instrument. Dis Esophagus. 2007;20(2):130–134. 10.1111/j.1442-2050.2007.00658.x [DOI] [PubMed] [Google Scholar]
- 15. Belafsky PC, Postma GN, Koufman JA. Validity and reliability of the reflux symptom index (RSI). J Voice. 2002;16(2):274–277. 10.1016/s0892-1997(02)00097-8 [DOI] [PubMed] [Google Scholar]
- 16. Yadlapati R, Kahrilas PJ, Fox MR, Bredenoord, AJ , Prakash Gyawali, C , Roman, S , et al., . Esophageal motility disorders on high‐resolution manometry: Chicago classification version 4.0©. Neuro Gastroenterol Motil. 2021;33(1):e14058, 10.1111/nmo.14058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Savarino E, Frazzoni M, Marabotto E, Zentilin P, Iovino P, Costantini M, et al. A SIGE‐SINGEM‐AIGO technical review on the clinical use of esophageal reflux monitoring. Dig Liver Dis. 2020;52(9):966–980. Epub 2020 Jun 6. PMID: 32513632. 10.1016/j.dld.2020.04.031 [DOI] [PubMed] [Google Scholar]
- 18. Gyawali CP, Rogers B, Frazzoni M, Savarino E, Roman S, Sifrim D. Inter‐reviewer variability in interpretation of pH‐impedance studies: the wingate consensus. Clin Gastroenterol Hepatol. 2021;19(9):1976–1978.e1. Epub 2020 Sep 2. PMID: 32890752. 10.1016/j.cgh.2020.09.002 [DOI] [PubMed] [Google Scholar]
- 19. Frazzoni M, Savarino E, de Bortoli N, Martinucci I, Furnari M, Frazzoni L, et al. Analyses of the post‐reflux swallow‐induced peristaltic wave index and nocturnal baseline impedance parameters increase the diagnostic yield of impedance‐pH monitoring of patients with reflux disease. Clin Gastroenterol Hepatol. 2016;14(1):40–46. 10.1016/j.cgh.2015.06.026 [DOI] [PubMed] [Google Scholar]
- 20. Li H, Johnson T. Wilcoxon's signed‐rank statistic: what null hypothesis and why it matters. Pharm Stat. 2014;13(5):281–285. Epub 2014 Jun 18. PMID: 24943680. 10.1002/pst.1628 [DOI] [PubMed] [Google Scholar]
- 21. Taylor M, Szafron B, Martin GP, Abah U, Smith M, Shackcloth M, et al. North west thoracic surgery collaborative (NWTSC). External validation of six existing multivariable clinical prediction models for short‐term mortality in patients undergoing lung resection. Eur J Cardio Thorac Surg. 2021;59(5):1030–1036. PMID: 33230562. 10.1093/ejcts/ezaa422 [DOI] [PubMed] [Google Scholar]
- 22. Paul P, Pennell ML, Lemeshow S. Standardizing the power of the Hosmer‐Lemeshow goodness of fit test in large data sets. Stat Med. 2013;32(1):67–80. Epub 2012 Jul 26. PMID: 22833304. 10.1002/sim.5525 [DOI] [PubMed] [Google Scholar]
- 23. Katzka DA, Pandolfino JE, Kahrilas PJ. Phenotypes of gastroesophageal reflux disease: where Rome, Lyon, and Montreal meet. Clin Gastroenterol Hepatol. 2020;18(4):767–776. 10.1016/j.cgh.2019.07.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information S1
Figure S1
Figure S2
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.