Address for Correspondence Michele Cicala MD, PhD, Department of Digestive Disease, Campus Bio Medico University, Via Alvaro del Portillo 200, 00128 Rome, Italy. Tel: +39 06 225411; fax: +39 06 22541456; e-mail: firstname.lastname@example.org
Background The sensitivity of 24-h pH monitoring is poor in non-erosive reflux disease (NERD). In NERD patients, the proximal extent of acid reflux is one of the main determinants of reflux perception. The present study was aimed to compare the diagnostic accuracy of acid exposure time (AET), at 5 cm above the lower esophageal sphincter, with those at 10 cm and at 3 cm below the upper esophageal sphincter as well as the reproducibility of these parameters.
Methods A total of 93 consecutive NERD patients, with typical symptoms responsive to proton pump inhibitor treatment, and 40 controls underwent esophageal manometry and multi-channel 24-h pH-test; 13 patients underwent the same study on two occasions. Symptom association probability (SAP) values were evaluated at each esophageal level.
Key Results The ROC curve indicates that the area under the curve was 0.79 at distal (SE = 0.039), 0.87 (SE = 0.032) at proximal (P = 0.029 vs distal), and 0.85 (SE = 0.033) at very proximal esophagus (P = 0.148). AET showed a reproducibility of 61% (Kappa 0.22) at distal esophagus, 77% (Kappa 0.45) at proximal and 53% (Kappa 0.05) at very proximal esophagus. The percentage of patients with a positive SAP was not significantly different when assessed at the distal compared with the proximal esophagus.
Conclusions & Inferences In NERD patients, the diagnostic yield of the pH test is significantly improved by the assessment of AET at the proximal esophagus. As this variable seems to be less affected by the day to day variability, it could be considered a reliable and useful diagnostic tool in NERD patients.
Among gastroesophageal reflux disease (GERD) patients, those not exhibiting esophageal mucosal injury at endoscopy, but a proven relationship between acid reflux and symptoms, namely non-erosive reflux disease (NERD) patients, account for nearly 70% and represent a heterogeneous group, according to the 24-h pH profile and to symptom-reflux association indices: indeed, it has been reported that in 30–50% of symptomatic patients, the results of currently available diagnostic techniques, i.e., endoscopy and pH-metry, remain within the normal range.1,2 In NERD patients, quality of life impairment is similar to that of erosive reflux disease (ERD) patients,3–5 moreover the response to standard acid suppressive treatments has been shown to be 20–30% lower than in patients with ERD. Therefore, in clinical practice, the management of NERD patients is still challenging. Indeed, the time-related variability of ambulatory pH monitoring findings is particularly high in NERD patients, in whom poor reproducibility of the findings has also been reported.6 Attempts have been made to improve the sensitivity of the ambulatory pH test. Increasing the duration of pH monitoring, more in keeping with physiological conditions, by means of wireless systems (Bravo), increases the likelihood of a significant reflux-symptom relationship and, as recently shown, significantly improves the reproducibility of the test.7,8 This technique, however, is hampered by the sampling rate of 6 s which might not change the overall esophageal acid exposure time (AET), but could potentially alter the correlation between symptoms and reflux episodes, and the capsule itself may even affect esophageal motility during the test.
Recently, a study aimed at comparing the diagnostic accuracy of pH monitoring at the very distal esophagus, 1 cm above the gastroesophageal junction (GEJ), with that at the traditional location (5 cm above GEJ), reported that measurement of very distal acid exposure improves the diagnostic accuracy as well as symptom-reflux correlation in erosive esophagitis, but not in NERD patients, in whom an improved diagnostic yield would be particularly helpful.9 Anggiansah et al. had shown a significantly reduced acid detection at 10 cm above the lower esophageal sphincter (LES) compared to 5 cm, however a Demeester score was used to define a pathological pH-test and, of interest, 3 out of the 14 patients with physiological acid exposure presented a higher score at the proximal, compared with the distal, esophagus.10
Growing evidence demonstrates that the proximal extent of acid reflux is relevant in triggering both typical and atypical symptoms. Weusten et al. were the first to suggest that, in GERD patients, the proximal extent of the refluxate is an important determinant for reflux perception.11 In an earlier study, we not only confirmed the higher frequency of proximal reflux in GERD patients but also reported an increased perception of proximal reflux episodes in NERD patients, particularly in those presenting normal AET at the distal esophagus.12 The relevance of proximal reflux has also been confirmed in an outcome study13 and in studies in which the pH-impedance technique was used.14–16 More recently, it has been demonstrated that the proximal extent of the refluxate is the factor most significantly associated with reflux perception in non-responder patients, during PPI treatment, thus confirming what had previously been shown in patients off therapy.17,18 The primary hypothesis of the present study was that the assessment of the acid exposure of the proximal esophagus better discriminates NERD patients from healthy control with respect to that of the distal esophagus. Therefore, in the present study, we compared, in NERD patients consecutively enrolled, the diagnostic accuracy of the AET values assessed at 5 cm (distal esophagus) above the LES, with those at 10 cm (proximal esophagus), and 3 cm (very proximal esophagus) below the upper esophageal sphincter (UES) as well as the reproducibility of these parameters.
Materials and methods
A total of 132 consecutive patients attending our outpatient unit for recurrent typical GERD symptoms – heartburn and/or acid regurgitation – lasting more than 6 months and responsive to PPI treatment (>60% improvement following the full PPI dose for at least 4 weeks), were invited to take part in the study. All patients underwent upper endoscopy, stationary esophageal manometry and multichannel ambulatory 24-h gastroesophageal pH monitoring, the latter performed between 3 and 14 days (median 5 days) after endoscopy.
Following endoscopy, 18 patients were excluded from the study due to duodeno-gastric peptic ulcer disease. Of the remaining 114 patients enrolled, 18 showing erosive esophagitis (ERD) (2 F, 16 M; mean age 45, SD 9 years), and 3 patients not presenting esophageal erosions, but evidence of erosive esophagitis at previous (3–5 years) endoscopy were also excluded. Our series of patients, therefore, comprised 93 NERD patients (51 F, 42 M; mean age 51, SD 8 years). The pH-metric data were compared with those of 40 asymptomatic, hospital staff volunteers (21 F, 19 M; mean age 46, SD 5 years) (healthy control group), all non-smokers. Patients and controls filled out a diary card on which position, time, and type of symptoms as well as times of meals, were reported. Clinical data were obtained using a standard structured questionnaire (Reflux Disease Questionnaire), which was filled out by all patients immediately prior to endoscopy, and included information concerning type, duration and frequency of symptoms. Patients on antisecretory, H2 antagonists and/or prokinetic drugs, stopped all treatment at least 3 weeks prior to endoscopy. None of the patients or controls in the study population had a history of gastrointestinal surgery, with the exception of appendectomy.
All NERD patients were invited to undergo a further pH study, 13 of whom agreed (5 F, mean age 42, SD 7 years), and the second study was performed after 2–4 weeks.
Written informed consent was obtained from all individuals and the study protocol was approved by the Ethics Committee of Campus Bio Medico University of Rome.
Stationary esophageal manometry
Esophageal manometry was performed with a perfused Dent-sleeve device assembly that incorporated a 6 cm sleeve sensor (Dentsleeve, Medtronic, Adelaide, Australia) and side-hole recording sites at 5, 10, and 15 cm above the sleeve. The sleeve catheter was passed transnasally and placed within the LES high-pressure zone. Low esophageal sphincter resting pressure was measured at the end-expiratory phase.
Multi-channel esophageal pH-metry
Ambulatory 24-h esophageal pH monitoring was performed using two probes, tied to each other, each with two antimony sensors, with a separate skin reference (Zinetics Medical Inc., Salt Lake City, UT, USA). Data were stored on a single portable digital recorder (Digitrapper pH200; Medtronic, Minneapolis, MN, USA). Before each study, the pH probes were calibrated in buffer solutions of pH 7 and pH 1. The four pH sensors were placed, according to the manometric findings, at gastric level, 5 cm above the LES, 10 and 3 cm below the UES.
A reflux episode was defined as a pH drop below 4 units at the distal esophageal sensor, lasting ≥4 s. AET was calculated at each esophageal site. Heartburn and acid regurgitation were considered in the analysis of symptoms. Reflux episodes were classified as symptom-related if they occurred ≤2 min before the onset of the symptom. The symptoms association probability (SAP) index was calculated according to the formula described elsewhere.19 AET was defined as pathological if the time, at pH <4, exceeded 4% of the total recording time.
The frequency distribution of AET was clearly asymmetric and the log-transformation reduced its asymmetry, providing a better fit to the gaussianity (according to Shapiro–Wilk statistics) as well as a reduction of the potentially detrimental effect of the outliers. The comparison between cases and controls was made by means of Student’s t-test with degrees of freedom (df), appropriately adjusted in case of variance heterogeneity (adj_df). In a secondary analysis where three groups were considered (healthy controls, NERD pH negative and NERD pH positive), anova was applied, followed by Tukey’s post hoc comparison to compare AET values of the two groups of patients vs the control group.
To compare diagnostic accuracy of AET measured at proximal and very proximal vs at distal esophagus level, ROC curves were obtained and compared by means of the procedure ROCCOMP available with the statistical software (STATA10, StataCorp; College Station, TX, USA). In addition, the pairwise comparisons between esophagus sites, in terms of sensitivity and specificity, were evaluated by means of McNemar test.
There was good compliance (no dropouts) to the procedure in the study population. The duration of pH-recording averaged 21.7 ± 0.1 h. AET values at each esophageal level in NERD patients compared to healthy controls are shown in Table 1. Mean values of AET were significantly higher in the whole group of patients at each esophageal site. When considering the subgroups of pH-positive and negative patients, according to the traditional cutoff related to the distal esophageal site, patients with a ‘physiological’ distal AET showed a significantly higher AET of the proximal esophagus compared with healthy controls.
Table 1. Acid exposure time, mean (95% CI), at different esophageal levels in healthy controls and NERD patients
NERD, non-erosive reflux disease.
*P < 0.01 vs healthy controls. †P = ns vs healthy controls.
Healthy controls (n = 40)
NERD (n = 93)
NERD pH positive (n = 53)
NERD pH negative (n = 40)
Despite the significant difference of the overall acid exposure time, the individual values showed an overlap between patients and controls, at each esophageal site, as shown in Fig. 1. The ROC analysis indicated that the area under the curve (AUC) of the distal AET was 0.79 (SE = 0.039), 0.87 (SE = 0.032) of the proximal AET (P = 0.029 vs distal), 0.85 (SE = 0.033) of the very proximal esophagus (p: n.s. vs distal) (Fig. 2).
The ROC procedure allowed us to define cutoffs which corresponded to specific levels of sensitivity or specificity. Assuming distal site as reference, we found that the traditional cutoff of 4% of 24-h acid exposure time resulted in a good specificity (88%) but in a unsatisfying sensitivity (57%). Looking at more proximal sites and requiring a specificity higher than 80% (thus setting at 20% the upper limit of false positives), we observed that the cutoffs of 1.0% at the proximal and 0.5% at the very proximal esophagus were more ‘discriminant’ than the 4% reference at distal site. Therefore, this approach identified the following cutoffs for AET: 4% at the distal, 1.0% at the proximal, and 0.5% at the very proximal esophagus.
The corresponding sensitivity and specificity are shown in Table 2, where the pairwise differences, are also reported. As shown in Table 2, sensitivity significantly increased (84% and 78%, vs 57%P < 0.001) while specificity slightly decreased (82% and 82%, vs 88%P = ns) when AET was investigated at the proximal and very proximal sites vs the distal site. The analysis of discordant cases is shown in the Table 2 (footnotes): a significantly higher proportion of patients showed values higher than cutoffs (true positive patients) when AET was assessed at the proximal and very proximal, vs the distal esophageal, site (for details see footnotes Table 2).
Table 2. Sensitivity and specificity of the abnormal pH values at the different esophageal levels
*50 true positive at both sites, 12 false negative at both sites, 31 discordant cases, 28 true positive only at middle, 3 true positive only at distal (McNemar, P = <0.001). †46 true positive at both sites, 3 false negative at both sites, 34 discordant cases, 27 true positive only at proximal, 7 true positive only at distal (McNemar, P = 0.001). ‡70 true positive at both sites, 12 false negative at both sites, 11 discordant cases, 3 true positive only at proximal, 8 true positive only at middle (McNemar, P = 0.227). §29 true negative at both sites, 1 false positive at both sites, 10 discordant cases, 6 false positive only at middle, 4 false positive only at distal (McNemar, P = 0.754). ¶29 true negative at both sites, 1 false positive at both sites, 10 discordant cases, 6 false positive only at proximal, 4 false positive only at distal (McNemar, P = 0.754). **30 true negative at both sites, 4 false positive at both sites, 6 discordant cases, 3 false positive only at proximal, 3 false positive only at middle (McNemar, P = 1.000).
Distal esophagus (cutoff 4%)
Proximal esophagus (cutoff 1.0%)
Very proximal esophagus (cutoff 0.5%)
Analysis of symptoms
Of the 93 NERD patients, 80 (47 with pathological AET at the distal esophagus, NERD pH+, and 33 with a physiological AET, NERD pH−) reported symptoms during the study day (mean 4.5, range 2–12). Heartburn accounted for 70%, regurgitation for 30% of symptoms. SAP values in patients are listed in Table 3. The percentage of patients with a positive SAP was higher, but not significantly different, at the distal, compared with the proximal esophagus. In particular, six patients with a positive SAP, at the distal esophagus, became negative when the SAP was measured at the proximal esophagus. Of these patients, three showed also a physiological AET at the proximal esophagus. On the other hand, four patients with a negative SAP index, at the distal site, became positive at the proximal esophagus. Of these patients, all but one, showed a physiological AET at the distal esophagus. The proportion of patients classified as SAP negative, at the proximal esophagus, was higher in NERD pH+ than in NERD pH− patients.
Table 3. SAP values assessed in patient groups at different esophageal levels
SAP, symptom association probability; NERD, non-erosive reflux disease.
NERD SAP positive, n (%)
NERD pH+ SAP positive, n (%)
NERD pH− SAP positive, n (%)
When considering NERD subgroups, according to AET and SAP at the distal esophagus, the statistical analysis performed excluding the 13 patients with normal AET and negative SAP confirmed the findings obtained in the entire sample, being sensitivity 66.2% at distal, 86.2% at proximal, and 77.5% at very proximal (respectively, P = 0.001 and P = 0.093 vs distal).
Moreover, also the AUC values assessed at the proximal and very proximal esophagus were higher than that assessed at the distal esophagus (0.88 and 0.85 vs 0.82).
In the 13 patients who underwent the same study on two separate occasions, the AET did not differ significantly between the two, at the three esophageal sites (Fig. 3).
AET showed a reproducibility of 61% (Kappa 0.22) at the distal esophagus (i.e., 61% of the tested subjects retained a normal or abnormal test result on both study days), 77% (Kappa 0.45) at the proximal and 53% (Kappa 0.05) at the very proximal esophagus. Even if the pH parameters at the proximal esophagus seem to be the most reproducible, the total percentage of time with a pH ≤4 showed a variation between the two study days by a factor of 3.0-fold (Fig. 3).
Results of the present study show that, in NERD patients, the diagnostic yield of the pH test is significantly improved by assessing AET at the proximal, compared with the distal, esophagus.
We focused on patients with recurrent and typical GERD symptoms, responsive to acid suppression, in the absence of mucosal injury at endoscopy. Efforts have been made to select ‘true’ NERD patients, i.e., those in whom previous endoscopic findings and/or an accurate washout from acid-suppressive treatment should have excluded a history of erosive disease. Over the last few decades, for as long as GERD continued to be identified mainly with the presence of esophagitis, and the diagnosis based upon endoscopy findings, attention was focused on the distal esophagus, 5 cm above the LES. Nevertheless, in NERD patients, the sensitivity of traditional ambulatory 24-h pH monitoring, in diagnosing and in offering an explanation for the disease, is poor (50–60%).6 As far as concerns pH monitoring, it is a conventional practice to first manometrically localize the proximal border of the LES and then to position the pH electrode, 5 cm proximal to this landmark.20 Historically, that position was chosen as a reasonable compromise: close enough to the squamo-columnar junction to detect gastroesophageal reflux and sufficiently distant to avoid migration into the stomach.21 However, no data exist yet confirming that the position 5 cm proximal to the LES is optimal for GERD diagnosis, particularly in the absence of mucosal injury. Results emerging from the present investigation show that the highest percentage of patients with a positive test, in terms of pathological AET, is obtained when considering the pH value at the proximal esophagus. Also findings from subjects classified discordantly, at the three esophageal levels, suggest that the higher sensitivity does not lead to a significant loss of specificity of the pH-metry when shifting the pH sensor from 5 to 10 cm above the LES. Moreover, comparison of the AUC of the pH measurements at the distal, proximal and very proximal esophagus, assessed by means of the ROC curve, indicates that the accuracy of the pH test significantly increases when considering the proximal AET.
According to the more recent definitions, NERD encompasses not only patients with evidence of a pathological acid exposure and/or a significant symptom-reflux association but also patients showing negative findings at the pH-test and symptoms relieved by acid suppression.4 In our series, in order to overcome possible criticisms, the statistical analysis was also performed after excluding the last subgroup of patients, in whom a placebo effect would have led to a selection bias and did not show different findings.
A recent study, aimed at assessing the accuracy of pH values in a group of GERD patients (erosive and NERD) at 1 cm above the LES, compared with the traditional 5 cm above the LES, failed to show an improvement in the diagnostic yield in the NERD subgroup.9 Growing evidence now supports the concept that, in the pathogenesis of NERD – and hence, of GERD symptoms – in addition to alterations in central pain processing of visceral stimuli, the spread of refluxate into the proximal esophagus plays a predominant role in eliciting typical symptoms.22
In the present study, when assessing the SAP index, at the distal and proximal esophagus, the proportion of patients with positive SAP did not differ, being significantly lower in the very proximal esophagus. We can offer no definite explanations for this finding. In the analysis of the association between symptoms and reflux, rigorous adherence to the threshold of pH 4, in the proximal esophagus, may lead to an underestimation of the reflux-related origin of the symptoms. Indeed, we have recently shown that a not negligible number (nearly 30%) of reflux episodes showed a decrease in acidity when spreading to the proximal esophagus.23 On the other hand, an improved accuracy of pH-impedance has been shown. It is well known that the SAP index values are related to the time windows chosen (ranging from 2 to 5 min) and the relationship between symptoms and reflux is made in a qualitative fashion. Although SAP values remain the best index to express the relationship between symptoms and reflux episodes,24 it is not uncommon that symptoms do not occur during the study day – nearly 15% in our series – or that for symptoms like hoarsness, globus, sore throat it is very difficult to discriminate onset and end, thus a quantitative measurement of pathological reflux would be helpful in clinical practice. Therefore, it is tempting to suggest that, shifting the pH sensor from 5 cm above LES to 10 cm below UES and/or having two pH sensors, one proximal and one distal, in the pH-impedance assembly, would achieve a more accurate characterization of the proximal reflux, and to further enhance the sensitivity of the test.
Our results, although emerging from a relatively small group of patients that agreed to undergo the pH monitoring on two separate occasions, also showed that the AET of the proximal esophagus appears to be the most reproducible. In particular, in our asymptomatic control group, the cutoff value was 1.0%; although these findings were derived from a relatively small population of healthy subjects, we are confident regarding the consistency of these data both on account of the very low variability in individual data, and the very similar results published in other series of asymptomatic controls.25
It is well known that the day-to-day variability in esophageal acid exposure is high and, as a result, the diagnostic reproducibility of 24-h pH measurement is low.6
In conclusion, in NERD patients, the diagnostic yield of the pH test is significantly improved by the assessment of AET at 10 cm below the UES compared to 5 cm above the LES. As this variable seems to be less affected by the day to day variability, it could be considered a reliable and useful diagnostic tool in non-erosive reflux disease.
Authors are grateful to Mrs Marian Shields for help with the English.
SE performed the study and wrote the manuscript; MR and PP analyzed the data; MC designed the research study and wrote the manuscript.