Introduction

According to the Rome III diagnostic criteria, evidence of a positive heartburn relationship with reflux, either by ambulatory reflux monitoring or through subjective outcome from antireflux treatment, is sufficient to incriminate gastroesophageal reflux disease (GERD) as the cause of the symptom [1]. In agreement with current guidelines, heartburn suppression by proton pump inhibitor (PPI) therapy is sufficient to diagnose GERD without further diagnostic testing in patients without alarm symptoms [2]. A substantial proportion of patients require lifelong daily PPI administration to control heartburn, and many of them ask for antireflux surgical or endoscopic interventions [3]: in such cases, further investigation is mandatory to objectively confirm the GERD diagnosis [2, 3]. As erosive reflux disease (ERD) is detected in a minority of patients with PPI-responsive heartburn, impedance–pH monitoring is warranted in most cases when a GERD diagnosis must be confirmed [4]. On impedance–pH monitoring, endoscopy-negative PPI-responsive patients are currently distinguished as having nonerosive reflux disease (NERD), defined by an abnormal esophageal acid exposure time (AET), or hypersensitive esophagus (HE), defined by a normal AET but positive symptom association probability (SAP) and/or symptom index (SI) [1, 3, 4]. On the other hand, functional heartburn (FH) is defined by the absence of evidence that reflux is the cause of the symptom; that is, unsatisfactory response to a PPI trial, negative endoscopy findings, normal AET, and negative SAP/SI [1, 3, 4]. Unfortunately, in clinical practice many PPI-responsive patients have a normal AET and do not perceive symptoms during 24-h impedance–pH testing or admit inaccurate symptom recording, theoretically requiring costly retesting. Moreover, SAP and SI are reportedly determined by chance occurrences when reflux rates are low [5]. Thus, the clinical value of SAP and SI appears at least questionable [6]. As PPI responsiveness is not considered sufficient to justify antireflux interventions [2, 3] and given the limits of SAP and SI, on impedance–pH monitoring many endoscopy-negative PPI-responsive patients can currently be distinguished as having NERD or HE only on the basis of an abnormal or a normal AET, respectively, leaving some diagnostic uncertainty in those with a normal AET [1–6]. Therefore, there is the need for objective impedance–pH criteria to confirm the diagnosis of HE before surgical or endoscopic antireflux procedures are considered.

Impedance–pH monitoring allows the most complete assessment of reflux: by detection of retrograde bolus transit, all reflux episodes independently of their acidity can be recorded [7]. Moreover, esophageal clearance of refluxate can be assessed [8]. Clearance of gastroesophageal reflux is biphasic. Volume clearance consists of a secondary peristaltic wave, which is elicited by esophageal stretch receptors and determines the end of a reflux episode removing around 90 % of the refluxate. Chemical clearance consists of a swallow-induced peristaltic wave, which is elicited by an esophagosalivary vagal reflex and delivers salivary bicarbonate and epidermal growth factor, thus augmenting the distal esophageal pH and favoring repair of reflux-induced mucosal damage. Impedance monitoring allows assessment of chemical clearance independently of volume clearance: a drop in impedance originating in the upper esophagus and reaching the lower esophagus signals peristaltic transit of saliva and has been defined as a postreflux swallow-induced peristaltic wave (PSPW) when it occurs within 30 s of the end of a reflux episode [9]. The PSPW index, obtained by division of the number of PSPWs by the number of reflux events, has been shown to efficiently separate ERD from NERD patients and both types of patients from non-GERD patients, not receiving as well as receiving PPI therapy [9]. Additionally, impedance monitoring allows assessment of baseline impedance: low values, unaffected by circadian variations, reflect reflux-induced impairment of mucosal integrity even in the absence of macroscopic damage [10–13]. The mean of three 10-min nighttime periods accurately reflects the 6-h nocturnal bedtime period, which is scarcely influenced by swallowing activity [14]. In endoscopy-negative heartburn patients with normal conventional impedance–pH findings, lower values of the mean nocturnal baseline impedance (MNBI) have been found in PPI responders as compared with PPI nonresponders [15], suggesting the inadequacy of conventional impedance–pH variables in separating HE from FH. Recently, the PSPW index and MNBI have been shown to increase the diagnostic yield of impedance–pH monitoring in patients with GERD as compared with healthy controls [16]. Likewise, these two novel impedance parameters could prove useful in distinguishing HE from FH.

The aim of the present study was to investigate whether impairment of chemical clearance, as assessed by the PSPW index, and of mucosal integrity, as assessed by MNBI, characterize HE as compared with FH during off-therapy impedance–pH monitoring independently of SAP and SI.

Methods

The study was designed and performed in accordance with the sixth revision of the Declaration of Helsinki. After protocol approval by institutional review boards, data collected from consecutive patients evaluated for clinical reasons at Italian referral centers between January 2012 and December 2014 were reviewed. All patients gave their signed informed consent. The inclusion criteria were as follows: (1) endoscopy-negative heartburn totally suppressed by PPI therapy, recurring after PPI withdrawal and again abolished by PPI therapy (PPI-responsive heartburn), in patients referred for possible antireflux surgical or endoscopic interventions; (2) endoscopy-negative heartburn unaffected by PPI therapy (i.e., totally refractory to 8-week high-dose PPI therapy) in patients with a normal AET and negative SAP/SI on impedance-pH monitoring. Symptoms of reflux and dyspepsia were assessed with a validated structured reflux–dyspepsia questionnaire based on a four-grade Likert-type scale scoring system [17, 18]. Impedance–pH monitoring was performed after 2 weeks of PPI withdrawal (antacids permitted) and was always preceded by esophageal manometry to exclude major motor disorders (i.e., achalasia and scleroderma esophagus) and for accurate location of the lower esophageal sphincter. The impedance–pH catheters adopted allowed the monitoring of changes in intraluminal impedance at 3, 5, 7, 9, 15, and 17 cm and in intraluminal pH at 5 cm, respectively, above the manometrically defined upper border of the lower esophageal sphincter. For the purpose of this study, exclusion criteria comprised Sjögren syndrome, previous esophagogastric surgery, and referral for regurgitation, chest pain, dysphagia, extraesophageal syndromes, or dyspeptic symptoms.

All impedance–pH tracings were blindly reviewed in a random order by expert observers (Marzio Frazzoni, Nicola de Bortoli, Manuele Furnari, Irene Martinucci, Salvatore Tolone, and Edoardo Savarino) rendered unaware of clinical details and the results of previous analyses. A dedicated software program (BioView Analysis; Sandhill Scientific, Highland Ranch, CO, USA) was used in conjunction with a 2 min time window visual analysis with zooming whenever this was deemed necessary. Data analysis was performed on liquid and liquid–gas reflux events for acidic (nadir pH < 4), weakly acidic (nadir pH between 4 and 7), and weakly alkaline (nadir pH not below 7) refluxes (mealtimes excluded). Conventional quantitative parameters included the period with esophageal pH < 4 (i.e., the percent AET) and the number of reflux events. According to our previous receiver operating characteristic (ROC) analysis in 50 nonoverweight healthy individuals eating their usual Mediterranean diet [16], cutoff values that maximized the sum of sensitivity and specificity (3.2 % and 48% respectively) were considered as threshold values. Positive SAP and SI were defined by rates of heartburn following reflux events of 95 % or more and 50 % or more respectively [19].

A PSPW was defined as an antegrade 50 % drop in impedance occurring within 30 s of a reflux event, originating in the most proximal impedance channel, reaching the most distal impedance channel, and followed by at least 50 % return to the baseline. Dividing the number of PSPWs by the number of reflux events, we obtained the PSPW index [9]. The MNBI was assessed from the most distal impedance channel during the nighttime recumbent period. Three 10-min periods (around 1:00 a.m., 2:00 a.m., and 3:00 am) were selected, and the mean was calculated to obtain the MNBI; periods including swallows, refluxes, and pH drops were avoided [14]. The threshold values for the PSPW index and MNBI were 61 % and 2292 Ω respectively [16].

According to conventional impedance–pH criteria [3, 4], the NERD group was defined on the basis of endoscopy-negative PPI-responsive heartburn with an abnormal AET, whereas the HE group was defined on the basis of endoscopy-negative PPI-responsive heartburn with a normal AET. Patients with endoscopy-negative heartburn unaffected by PPI therapy (i.e., totally refractory to 8-week high-dose PPI therapy) and with a normal AET and negative SAP/SI constituted the FH group.

Statistics

Differences among the three study groups were examined by analysis of variance for continuous variables and the chi-square test for categorical variables, with Bonferroni correction for multiple comparisons and significance set at P < 0.017.

The diagnostic accuracy of quantitative impedance parameters—namely, the number of reflux events, the PSPW index, and the MNBI—was assessed. Sensitivity, specificity, and positive and negative predictive values with 95 % confidence intervals (CI) were calculated on the basis of best cutoff values previously established in healthy controls on ROC analysis [16]. Then, focusing on HE, we performed a multivariate logistic regression analysis adjusted for age, sex, body mass index, and presence of hiatal hernia to establish whether the PSPW index and MNBI were independent predictors of HE as compared with FH. The efficacy of the PSPW index and MNBI in separating HE from FH was then assessed by means of ROC analysis with calculation of the area under the curve (AUC) and the 95 % CI. Analyses were performed by Leonardo Frazzoni and Andrea Farioli with STATA, release 13 (StataCorp, College Station, TX, USA).

Results

Records from 125 NERD, 108 HE, and 70 FH patients were reviewed. No adverse event had been registered during impedance–pH testing and all 303 impedance–pH tracings were blindly reanalyzed. The PSPW index and the MNBI could be calculated in all tracings. The main baseline characteristics and impedance–pH findings of patients are reported in Table 1. Notably, the PSPW index and MNBI were significantly lower in NERD than in HE and in both of them as compared with FH.

Table 1 Baseline characteristics of patients with nonerosive reflux disease (NERD), hypersensitive esophagus (HE), and functional heartburn (FH)
Full size table

Table 2 shows the diagnostic accuracy of impedance parameters in NERD and HE. Higher sensitivity (no overlap of 95 % CI) was observed in a comparison of the PSPW index and MNBI with the number of reflux events in both NERD and HE, as well as in a comparison of the PSPW index with the MNBI. High specificity was observed with all three quantitative impedance parameters. On multivariate logistic regression analysis, the PSPW index (adjusted odds ratio 0.863, 95 % CI 0.819–0.908, P = 0.001) and the MNBI (adjusted odds ratio 0.998, 95 % CI 0.997–0.999, P = 0.001) were independent predictors of HE. On ROC analysis, combined assessment of the PSPW index and the MNBI allowed excellent separation of HE from FH: the AUC was 0.957 (95 % CI 0.926–0.991), significantly greater than that obtained with the PSPW index (0.924, 95 % CI 0.879–0.969, P = 0.0254) and the MNBI (0.864, 95 % CI 0.809–0.919, P = 0.0006) separately (Fig. 1).

Table 2 Diagnostic accuracy of quantitative impedance parameters in nonerosive reflux disease (NERD) and in hypersensitive esophagus (HE) as compared with functional heartburn (FH)
Full size table
Fig. 1
figure 1

Receiver operating characteristic (ROC) curves for hypersensitive esophagus versus functional heartburn. On ROC analysis, combined assessment of the postreflux swallow-induced peristaltic wave (PSPW) index and the mean nocturnal baseline impedance (MNBI) afforded an area under the curve (0.957, 95 % CI 0.926–0.991, black curve) significantly greater than that obtained with the PSPW index (0,924, 95 % confidence interval 0.879–0.969, P = 0.0254) and the MNBI (0.864, 95 % confidence interval 0.809–0.919, P = 0.0006) (gray curves) separately

Full size image

Table 3 reports the diagnosis of HE in 108 patients as confirmed by SAP and SI or by the PSPW index and MNBI. Overall, positive SAP and/or SI was found in 67 of the 108 HE patients (62 %), whereas an abnormal PSPW index and/or MNBI was found in 99 of the 108 HE patients (92 %; P < 0.0001; Fig. 2).

Table 3 Symptom association probability (SAP) and symptom index (SI) positivity as compared with postreflux swallow-induce peristaltic wave (PSPW) index and mean nocturnal baseline impedance (MNBI) positivity in 108 patients with hypersensitive esophagus
Full size table
Fig. 2
figure 2

Positive symptom association probability (SAP) and/or symptom index (SI) versus abnormal postreflux swallow-induce peristaltic wave (PSPW) index and/or mean nocturnal baseline impedance (MNBI) in 108 patients with hypersensitive esophagus. SAP and/or SI positivity in 67 of the 108 patients (62 %) and abnormal PSPW index and/or MNBI in 99 of the 108 patients (92 %; P < 0.0001, chi-square test). The threshold values for the PSPW index and MNBI were 61 % and 2292 Ω respectively

Full size image

Discussion

In uninvestigated patients with heartburn, symptom response to PPI therapy confirms GERD [1, 2]. Currently, in patients with PPI-responsive heartburn NERD is defined by negative endoscopy findings and an abnormal AET [3, 4]. Since the introduction of the Rome III diagnostic criteria [1], HE has been included within the realm of GERD and is currently defined by negative endoscopy findings and normal AET but positive SAP/SI and positive response to PPIs [3, 4]. On the other hand, FH has been defined by the absence of evidence that reflux is the cause of the symptom (i.e., PPI refractoriness, negative endoscopy findings, normal AET, and negative SAP/SI) [1, 3, 4]. Given the limits of symptom-reflux association indexes, we aimed to establish whether the PSPW index and MNBI are predictors of HE independently of SAP and SI. For this purpose, we blindly reviewed off-therapy impedance–pH tracings from 233 consecutive patients with endoscopy-negative, PPI-responsive GERD, subdivided into 125 NERD and 108 HE cases on the basis of abnormal or normal AETs respectively, and from 70 consecutive FH patients defined as above. Remarkably, the PSPW index and MNBI were significantly lower in NERD than in HE and in both of them as compared with FH, confirming the pathophysiologic soundness of these two novel impedance parameters [16]. The PSPW index and MNBI were very efficient in separating HE from FH, significantly more so than SAP and SI, and were abnormal in the vast majority of cases with negative or discordant SAP/SI. The PSPW index and MNBI proved to be independent predictors of HE on multivariate analysis, and their combined assessment afforded an excellent AUC on ROC analysis.

In clinical practice, impedance–pH monitoring is mainly used to confirm GERD in endoscopy-negative patients with PPI-responsive heartburn before antireflux surgical or endoscopic interventions, or to establish whether PPI-refractory reflux symptoms in endoscopy-negative patients are reflux related or reflux unrelated [2, 3]. Conventional assessment of impedance–pH tracings includes AET and SAP/SI evaluation. AET is regarded as the most useful single discriminator between physiologic and pathologic reflux but it may be normal in up to 19 % of patients with ERD [16, 19]. In the present series, in nearly half of GERD patients the AET was normal, a result confirming the moderate sensitivity of this parameter in endoscopy-negative patients [16], possibly due to its day-to-day variability [20]. In patients with endoscopy-negative heartburn and a normal AET, SAP and SI have been the symptom–reflux association indexes most commonly used to separate HE from FH [5, 14–16, 21–26]. However, the clinical value of SAP and SI has recently been questioned [6], particularly when discordant results are found [3] and when reflux rates are low [5]. Moreover, in clinical practice patients often do not have symptoms or report one or two symptom episodes only during 24-h impedance–pH testing, or admit inaccurate symptom recording, mostly during nighttime and when multiple symptoms are assessed. Finally, symptom–reflux association indexes do not measure the severity and clinical relevance of symptoms: for instance, sleep interruption due to nocturnal heartburn is undoubtedly troublesome, but SAP and SI may nonetheless be negative because they are calculated for the entire monitoring period. In our series, concordant SAP/SI positivity was observed in only 46 % of HE patients as defined by PPI responsiveness and a normal AET, and discordant positivity was found in 16 % of cases. These results are in accordance with the view that diagnosis of HE based on SAP/SI positivity is insufficiently accurate and reliable for clinical decisions [6], in particular for suggesting or advising against antireflux endoscopic or surgical interventions.

By impedance monitoring, the number of reflux events independent of pH can be recorded but has more commonly been used to investigate patients with reflux symptoms during on-therapy testing [27–30] than during off-therapy testing [24] and is not included in the current definitions of NERD and HE [3, 4]. In this study we confirmed the diagnostic limits of this parameter [16], which showed good specificity but poor sensitivity in separating HE from FH. By representing bolus transit, impedance–pH monitoring allows also assessment of retention of refluxate until volume clearance, elicited by a secondary peristaltic wave, occurs determining the end of the reflux episode [7, 8]. Volume clearance can be assessed by the bolus clearance time but we did not consider this parameter owing to lower sensitivity than the number of reflux events [16] and because the accuracy of automated software calculation is significantly and substantially lower than manual calculation, the latter being time-consuming and thus unsuitable for clinical practice [31].

Postreflux salivary swallows are elicited by acidic as well as weakly acidic refluxes [9], the latter representing the vast majority of reflux events during antisecretory therapy [32]. The PSPW index is therefore suitable for evaluation of patients not receiving PPI therapy as well as for patients receiving PPI therapy, with lower values in ERD than in NERD patients and in both groups as compared with non-GERD subjects [9]. Baseline impedance reflects mucosal integrity [10–13], lower values distinguishing PPI-responsive from PPI-refractory heartburn patients with normal conventional impedance–pH variables [15] and associated with greater probability of PPI response in patients with chronic cough [33]. Recently, we showed that analyses of the PSPW index and MNBI increase the diagnostic yield of impedance–pH monitoring in GERD patients as compared with healthy controls [16]. In clinical practice, however, the key issue is whether the PSPW index and MNBI can distinguish reflux-related from reflux-unrelated heartburn in patients with a normal AET [34], as there is little doubt that patients with heartburn and an abnormal AET have GERD. Therefore, we focused on HE in comparison with FH. A gold standard for GERD diagnosis is still lacking [23] but heartburn suppression with PPIs is considered as direct evidence of GERD [1, 2]. Accordingly, and taking into account the limits of SAP and SI, we considered unambiguous heartburn responsiveness to PPIs in endoscopy-negative patients with a normal AET as evidence of HE irrespective of SAP/SI. For the purpose of this study, we considered patients as PPI responsive when heartburn was suppressed with PPI therapy, recurred after PPI withdrawal, and was again abolished with PPI therapy: it was not just a PPI trial, which is limited by moderate specificity [35]. We compared HE patients with patients with a firm diagnosis of FH (endoscopy-negative heartburn unaffected by PPI therapy, i.e., totally refractory to PPI therapy in addition to a normal AET and negative SAP/SI) because we aimed to establish whether HE is characterized by impairment of chemical clearance and mucosal integrity during off-therapy impedance–pH monitoring in comparison with non-GERD patients with the same symptom (i.e., heartburn). Therefore, we did not consider endoscopy-negative patients with PPI-refractory heartburn and an abnormal AET and/or positive SAP/SI because the suspicion of refractory GERD in these cases is high and more detailed investigations, including on-therapy impedance–pH monitoring, have been recommended [2, 3] to clarify the mechanism(s) of PPI refractoriness before antireflux interventions are advised.

In patients with HE, we found that the PSPW index and MNBI were more sensitive than the number of reflux events, the PSPW index being more sensitive than the MNBI. Multivariate analysis showed that the PSPW index and MNBI were independent predictors of HE, suggesting that both parameters should be analyzed. Accordingly, combined assessment of the PSPW index and MNBI allowed effective separation of HE from FH on ROC analysis, affording an excellent AUC (more than 0.90), significantly greater than that achieved by assessment of the two parameters separately. Notably, the PSPW index and MNBI were abnormal in the vast majority of HE patients with negative or discordant SAP and SI, who represented more than half of the patients in our series, and HE diagnosis was confirmed significantly more often with an abnormal PSPW index and/or MNBI than with concordant and discordant SAP/SI positivity. These results show that HE diagnosis can be confirmed in the vast majority of cases with combined analysis of the PSPW index and MNBI independently of SAP and SI.

Impairment of chemical clearance and of mucosal integrity can explain increased perception of reflux events in HE. Impairment of chemical clearance, as expressed by a low PSPW index, implies prolonged contact of esophageal mucosa with acidic and weakly acidic refluxes [9] and inhibition of mucosal reparative processes, playing a relevant role in persistence of esophageal mucosal breaks [36] and in neoplastic progression of esophageal intestinal metaplasia [37] despite adequate acid suppression. Low baseline impedance reflects impaired mucosal integrity [10–13], in turn reflecting dilated intercellular spaces, which favor backward diffusion of refluxate; moreover, low baseline impedance is directly related to heartburn intensity [12, 13] and to PPI-induced heartburn suppression [15]. Therefore, impairment of chemical clearance and of mucosal integrity can well promote reflux perception and trigger reflux-related heartburn, explaining hypersensitivity of the esophageal mucosa to apparently physiological reflux burden as expressed by a normal AET and a normal number of reflux events.

A few limitations of our study should be acknowledged. Our study concerns a retrospective review of highly selected cases evaluated for clinical reasons at tertiary care centers; however, the data were prospectively collected by our adopting standardized criteria, and impedance–pH tracings were blindly reviewed by expert observers rendered unaware of clinical details and the results of previous analyses to ensure objective evaluation. The threshold values for the AET and the number of reflux events adopted in our study are lower than the values usually adopted: our normative values were defined as the best cutoff on ROC analysis comparing 289 GERD patients with 50 nonoverweight healthy controls eating a Mediterranean diet [16] and are quite similar to upper interquartile values recently reported by Zerbib et al. [38] in 46 healthy volunteers. Finally, although there is some evidence that SAP/SI positivity can predict positive surgical outcome [28, 30, 39, 40], prospective outcome studies are warranted to define the predictive value of the PSPW index and MNBI in patients considered for antireflux surgical or endoscopic interventions.

Currently, despite being time-consuming, visual analysis of impedance–pH tracings is mandatory because automated software analysis is not accurate enough [41]. The PSPW index and MNBI appear to be highly suitable because their calculation requires only a few additional minutes during visual inspection of tracings and interobserver reproducibility is very high [16]. Moreover, we confirm their high applicability [16] even with low reflux rates as occurring in patients with HE and FH. The PSPW index and MNBI appear particularly useful when GERD diagnosis is in doubt (i.e., when the AET is normal and SAP and SI are negative or discordant, or the patient admits poor accuracy in symptom recording).

In conclusion, HE is characterized by impairment of chemical clearance and of mucosal integrity, which explains increased perception of reflux events. Combined assessment of the PSPW index and MNBI allows objective diagnosis of HE independently of and significantly more often than SAP and SI. In clinical practice, when SAP and SI afford uncertain results, the PSPW index and MNBI should be analyzed so that GERD diagnosis is not overlooked.