Comprehensive analysis of acidic pharyngeal reflux before and after proton pump inhibitor treatment in patients with suspected laryngopharyngeal reflux

Abstract

Objectives

The usefulness of pharyngeal pH monitoring in patients with symptoms attributed to laryngopharyngeal reflux (LPR) has been questioned. One problem is the uncertainty whether the pharyngeal pH monitoring captures the aspects of LPR which are responsible for symptoms. We aimed to gain more insight into this problem by performing a comprehensive analysis of acidic pharyngeal reflux before and after the treatment with proton pump inhibitors (PPIs) in patients with suspected LPR.

Methods

We used simultaneous pharyngeal and distal esophageal 24-hour pH/impedance monitoring to establish the gastroesophageal origin of pharyngeal reflux, and an unbiased approach to analysis by evaluating a whole range of pharyngeal reflux acidity (pH < 6, pH < 5.5, pH < 5.0, pH < 4.5 and pH < 4.0).

Results

PPI treatment substantially (by ~50%) improved the symptoms attributed to LPR. In contrast, PPI did not reduce the number of pharyngeal reflux episodes or duration of pharyngeal acid exposure at any pH level. This was also true in a subgroup of patients considered to be good responders to PPI (symptoms improvement by ~75%). Furthermore, good responders did not have more acidic pharyngeal reflux than the patients who were less responsive to PPI.

Conclusions

PPI treatment did not reduce acidic pharyngeal reflux despite substantially improving the symptoms attributed to LPR. This may be because pharyngeal pH monitoring does not quantitatively capture the aspects of LPR responsible for symptoms or because acid causes the symptoms also by mechanisms other than LPR. Our results argue against the utility of pharyngeal pH monitoring in patients with suspected LPR.

Introduction

Laryngopharyngeal reflux (LPR) is suspected when the symptoms, typically hoarseness, throat clearing, globus and cough, are attributed to gastroesophageal reflux reaching the mucosa of the upper airways. Proton pump inhibitors (PPIs) are the mainstay of treatment of suspected LPR, and the need for sufficient duration of treatment and the use of high doses of PPI is emphasized [1,2]. The PPI treatment has been found effective to improve the symptoms and laryngeal injury attributed to LPR in some patients, but only partially effective or ineffective in others [3-6]. The effect of PPI on symptoms and laryngeal injury attributed to LPR is thought to be due to inhibition of production of gastric acid as the key corrosive component of LPR.

The utility of monitoring of acidic LPR by various techniques has been questioned. Increasing number of studies suggests that the monitoring of LPR has limited role in the diagnosis of LPR and in the prediction of the response to the PPI treatment [7-10]. However, at least some studies arrived to different conclusions [11-13]. In any case, it has been increasingly recognized that the evaluation of the utility of LPR monitoring is often complicated by several confounding factors.

First, the need for repeated LPR monitoring before and after PPI treatment adds a significant logistic burden as patients are often unwilling to undergo a second 24-hour monitoring after PPI treatment, especially if their symptoms are improved. In addition, because prior PPI treatment can confound the analysis, the patients need to be off PPI therapy at the time of enrollment. It is difficult to enroll such patients because of a widespread use of PPI [7,9,14]. Second, it is unknown how much acid is required to cause symptoms and laryngeal injury. Therefore, a comprehensive analysis of wide range of acidity (pH) of pharyngeal reflux is needed. Third, the evaluation of pharyngeal pH without monitoring the esophageal pH and impedance may introduce further variability because the pH drop in the pharynx due to gastroesophageal reflux (GER) cannot be reliably differentiated from the pH drop due to swallowing of food, liquids or saliva.

Here, we tried to limit the effect of these factors. The aim of this study was to determine whether the PPI treatment reduces pharyngeal acidic reflux similarly as it reduces the symptoms attributed to LPR. We enrolled patients who were off PPI therapy for at least 30 days. We used dual simultaneous pharyngeal and distal esophageal pH/impedance monitoring with individually selected spacing of pH sensors to monitor pH in the pharynx and esophagus according to high-resolution manometry (HRM). We performed this monitoring before and after 12 weeks of PPI treatment. Finally, we used our recently developed approach for comprehensive analysis of the pharyngeal reflux episodes (acidity, number and duration) [15]. Our data indicate that despite substantial improvement of symptoms and laryngeal injury after PPI treatment, no substantial changes in pharyngeal acidic reflux were detected on PPI. Our data suggest that 24-hour pharyngeal and esophageal pH/impedance monitoring in patients diagnosed with LPR is of limited clinical usefulness.

Patients and methods

The protocol of the study was approved by the Ethics Committee of Jessenius Faculty of Medicine, Comenius University with the approval number EK 1485/2014. This was a prospective single-center study which is a continuation of our previously published study [15]. The evaluation was performed on outpatient basis. Consecutive patients referred for suspected LPR to Gastroenterology Clinic, Jessenius Faculty of Medicine between November 2014 and May 2016 were prospectively screened for eligibility. A careful interview was conducted to assess the symptoms suggestive of LPR and personal history was obtained. Inclusion criteria were laryngopharyngeal symptoms such as cough, hoarseness, globus and throat clearing for >6 months. Exclusion criteria were age below 18 years old, infection of the upper or lower airways in the previous month, smoking, alcohol consumption >40 g/day, history of thoracic or abdominal surgery, pregnancy or breast feeding, neurological disorders, active malignancy, inflammatory bowel disease and presence of gastric inlet patch in the esophagus on the upper endoscopy. Written informed consent was obtained from each eligible subject who agreed to be enrolled into the study. Thirty-five subjects who met the initial inclusion criteria were evaluated. These subjects were instructed to withhold PPI therapy for 30 days. After this period, the subjects completed reflux symptom index (RSI) questionnaire [16], underwent HRM, simultaneous pharyngeal and esophageal 24-hour pH/impedance monitoring (as described in detail below), and flexible laryngoscopy during which the reflux finding score (RFS) [17] was obtained by using the 3.5-mm rhino-pharyngo-laryngofibroscope 11101 RP2 (Karl STORZ, Tuttlingen, Germany). RFS was obtained by an experienced ENT physician (T.H.). Only the subjects that had positive RSI (RSI > 13) or RFS (RFS > 7) and at least one LPR episode on 24-hour pH/impedance monitoring (as defined below) were further investigated (N = 27). The remaining patients had either RSI ≤13 and RFS ≤7 (N = 2), no LPR episode on 24-hour pH/impedance monitoring (N = 5), or technical issues during 24-hour pH/impedance monitoring that made evaluation of the recording impossible (N = 1). The 27 investigated patients were prescribed PPI treatment twice daily and were carefully instructed to take PPI regularly, approximately 30 minutes before the breakfast and dinner. Most patients received pantoprazol 2 × 40 mg (N = 9) or rabeprazol 2 × 20 mg (N = 5), while others received omeprazol 2 × 20 mg (N = 2), lanzoprazol 2 × 30 mg (N = 1) and esomeprazol 2 × 40 mg (N = 1). These dosages are the equivalent of recommended PPI dosing for the treatment of laryngopharyngeal reflux. The patients were clearly instructed not to take any other medication that could potentially affect LPR, and the use of other medication (if any) was reviewed during the follow-up visit. After 12 weeks of the PPI treatment, the patient was invited for a follow-up visit where careful interview was carried out and compliance with the medication was assessed. RSI and RFS were again obtained, and 24-hour simultaneous pharyngeal and esophageal pH/impedance was again performed this time while still on PPI treatment. Of 28 enrolled patients who were prescribed PPI, 18 patients completed the study and were included in the final analysis. Of the remaining patients, three patients did not take PPI according to the recommendation, two patients came to the second visit, but had acute upper airways infection that made the evaluation of RSI and 24-hour pH/impedance impossible. One patient was unavailable due to travel abroad for work. We have no information about the remaining four patients.

HRM and simultaneous pharyngeal and esophageal 24-hour pH/impedance monitoring were performed after overnight fast. We used ManoScan ESO High Resolution Manometry System (Given Imaging). Upper esophageal sphincter (UES) and lower esophageal sphincter (LES) were located. We adopted widely accepted approach of positioning the proximal pH probe 1 cm above the UES and the distal pH probe 5 cm above the LES determined by HRM [7,9,18]. In our view, simultaneous pharyngeal and distal esophageal pH/impedance monitoring has the best potential to detect LPR as it allows tracing the progress of a reflux episode from the distal esophagus toward the pharynx. Simultaneous pharyngeal and esophageal 24-hour pH/impedance monitoring was performed with pH/impedance catheters VersaFlex with two pH sensors and eight impedance electrodes (given imaging). Three sizes of catheters with the distance of 15, 19 and 22 cm between pH sensors were available for the study. In catheters with pH sensors distance 19 and 22 cm, impedance electrodes were 1 cm above and 1 and 3 cm below the distal pH sensor, 1 cm above and 2, 5 and 7 cm below the proximal pH sensor (Fig. 1a). In catheters with pH sensors distance 15 cm, impedance electrodes were 1 and 3 cm below and 1, 3 and 5 cm above the distal pH sensor and 1 and 3 cm below and 1 cm above the proximal pH sensor. Before recording, the catheters were calibrated using pH 4.0 and 7.0 buffer solutions. The appropriate catheter size was selected according to the locations of the UES and LES determined by HRM so that when the catheter was placed transnasally the proximal pH sensor was approximately 1 cm above the proximal margin of the UES and the distal pH sensor was approximately 5 cm above the proximal margin of the LES. If this was not possible to achieve, the catheter of the smaller diameter between the pH sensors was selected and placed according to the UES location. The average position of the distal pH sensor was 6 cm above the proximal margin of the LES (i.e., in the majority of patients the distal probe was positioned between 5 and 6 cm above the proximal margin of the LES). Patients were encouraged to maintain their normal activities, sleep schedule, and eat their usual meals during the 24-hour monitoring. We used Digitrapper pH-Z Testing System for the 24-hour recording. Patients were also asked to mark the time of eating and the horizontal body position on the data recorder. When performing pH/impedance while ‘on PPI', the same catheter type has been chosen as in the first measurement, also attention has been paid to achieve the same position of the pH sensors. Moreover, patients were instructed to mark the time of taking PPI on the data recorder.

Fig. 1.

Recording of pharyngeal reflux. (a) Representative example of a dual pH sensor catheter placing. The pH sensors were located 1 cm above the UES and 5 cm above the LES. Impedance electrodes are located 1 cm above and 2, 5 and 7 cm below the proximal pH sensor and 1 cm above and 1 and 3 cm below the distal pH sensor. (b) Representative trace of pharyngeal acidic reflux episode recorded by simultaneous pharyngeal and distal esophageal pH/impedance. Sudden drop of pH and impedance in the distal esophagus (black arrow, GER) is followed by distal to proximal decrease in impedance (distal to proximal movement of the refluxate) and subsequent drop of pH in the pharyngeal pH sensor (red arrow, pharyngeal reflux). GER, gastroesophageal reflux; LES, lower esophageal sphincter; UES, Upper esophageal sphincter..

Tracings were visually analyzed using AccuView pH-Z version 5.2 software (Given Imaging). One blinded observer was responsible for evaluation of pH/impedance studies. Meals were excluded from the analysis. Gastroesophageal reflux events by impedance changes were defined on the basis of previous human studies [19,20]. Liquid gastroesophageal reflux was defined as a retrograde 50% drop in impedance starting distally and propagating to at least the next two more proximal measuring segments. Gas gastroesophageal reflux was defined as simultaneous increase in impedance above 5000Ω starting in the most distal impedance channels and propagating to at least the next two more proximal impedance measuring segments. Mixed gastroesophageal reflux (liquid–gas) was defined as gas reflux occurring simultaneously before or during a liquid reflux.

Consistent with the studies in the field, we used the pharyngeal reflux as a surrogate of laryngopharyngeal reflux [18,21-25]. Pharyngeal reflux was identified when a gastroesophageal reflux episode (as defined above) was temporarily associated with a drop of pH in the proximal pH sensor (Fig. 1b). Special effort was dedicated to exclude pH changes in the proximal pH sensor due to swallowing, sensor drying and other artifacts based on their know pH/impedance patterns. We based our analysis on the assumption that in the absence of swallows a sudden drop of pH in the hypopharynx occurring in a close temporal sequence with a preceding reflux detected by distal pH/impedance is a pharyngeal reflux event. In order to allow for comprehensive analysis of LPR, we performed a separate analysis for proximal pH ≤6, 5.5, 5, 4.5 and 4. If pH in the proximal sensor was already lower before the reflux event, the reflux was considered as pharyngeal reflux when pH drop of at least 0.5 unit was observed during the reflux. Pharyngeal reflux was quantified as the number of pharyngeal reflux events for each pH. The pharyngeal acid exposure time for each pH was calculated as the cumulative time of pH ≤6, 5.5, 5, 4.5 and 4. For example, the duration of pharyngeal acid (pH < 5.0) exposure was calculated as cumulative time pH of pharyngeal reflux was pH <5.0 during 24 hours.

In addition, tracings were also reviewed for the number of pharyngeal reflux according to the criteria developed previously. According to Zerbib et al. [18], LPR was defined as a retrograde 50% impedance drop, both distally and proximally, within the entire esophagus if no swallow occurred during the pharyngeal impedance drop. According to Kawamura et al. [26], pharyngeal reflux events were defined analogously to gastroesophageal reflux events. A liquid reflux event was defined as an impedance drop of at least 50% from baseline and below 1200Ω preceded by or simultaneous with an esophageal liquid reflux event [26]. The esophageal acid exposure time and DeMeester score were automatically calculated by the analysis software. Furthermore, the number of all reflux events, number of liquid, gas and mixed reflux events and number of acidic and weakly acidic reflux events were obtained.

Statistical analysis

The number of reflux episodes is expressed as median and interquartile range (IQR). The duration of reflux episodes is expressed as mean ± SEM. Paired t-test, unpaired t-test and two-way repeated measures analysis of variance (ANOVA) with Sidak's post-hoc multiple comparisons test were used as indicated in the text and figure legends.

Results

Eighteen patients (eight females/10 males, average age 48.7 ± 10.3 years, range 33–70 years) completed the study. The symptoms (RSI), laryngeal injury (RFS) and pharyngeal reflux were evaluated before and after 12 weeks of PPI treatment. The PPI treatment improved the symptoms and laryngeal injury attributed to LPR. In the whole group of patients the LPR symptom index RSI was reduced by more than 50% from median 24 to median 11 (Fig. 2a). The laryngeal pathology attributable to LPR and quantified as RFS was also substantially reduced by PPI treatment. RFS before and after PPI treatment obtained in 15 patients was 10 [9–12] and 6 [4–7], respectively (P < 0.01, paired t-test). As described in methods, the physicians performing RFS assessment after the PPI treatment were unaware of the results of pH/impedance studies, but they were aware of the PPI treatment. Therefore, we report RFS with the caveat that it was obtained by the physicians not blinded to the PPI treatment.

Fig. 2.

PPI treatment improved symptoms attributed to LPR, but did not reduce pharyngeal reflux at any pH level. (a) RSI before and after PPI treatment. **P < 0.01, paired t-test, N = 18. (b) Duration of pharyngeal acid exposure before and after PPI treatment. The duration of pharyngeal acid exposure for each pH level was calculated as a cumulative time during which the pH of pharyngeal reflux was below the indicated level during 24 hours (two-way repeated measure ANOVA, P = 0.25, N = 18). (c) Number of pharyngeal acidic reflux episodes with indicated acidity during 24 hours before and after PPI treatment (two-way repeated measure ANOVA followed by Sidak's multiple comparisons test, *P < 0.05, N = 18). ANOVA, analysis of variance; LPR, laryngopharyngeal reflux; PPI, proton pump inhibitor; RSI, reflux symptom index.

In contrast to substantially reducing the symptoms (RSI) and the laryngeal injury (RFS) attributed to LPR, the PPI treatment did not inhibit pharyngeal reflux. We used simultaneous pharyngeal and distal esophageal 24-hour pH/impedance monitoring to detect reflux from the esophagus into the pharynx (esophago-pharyngeal reflux) denoted here as pharyngeal reflux. Preceding distal GER was required to consider pharyngeal drop in pH as pharyngeal reflux. In order to evaluate pharyngeal reflux independently of the assumption of how acidic the pH of pharyngeal reflux is required to cause laryngeal injury and the symptoms of LPR, we performed a stratified analysis for different levels of pH of pharyngeal reflux (pH < 4.0, pH < 4.5, pH < 5.0, pH < 5.5 and pH < 6.0). This approach to analysis is discussed in detail in our previous study [15].

We found no decrease in acidic pharyngeal reflux on PPI at any level of pH evaluated (Fig. 2b and c). Neither the duration of pharyngeal acid exposure nor the number of pharyngeal reflux episodes was decreased (Fig. 2b and c). In fact, the number of pharyngeal reflux episodes with pH <5.5 and pH <6.0 was slightly increased. We also evaluated pharyngeal reflux as defined by impedance-based criteria and found that it was not reduced on PPI treatment. When using impedance-based criteria developed by Zerbib et al. [18], the number of pharyngeal reflux episodes was 1 [1–4] and 1 [0–1], before and after PPI treatment, respectively (P = 0.3, paired t-test). When using impedance-based criteria by Kawamura et al. [26], the number of pharyngeal reflux episodes was 1 [0–2] vs 0 [0–0] (P = 0.2, paired t-test). Thus, despite the reduction of symptoms and laryngeal injury by PPI treatment, we found no reduction of pharyngeal reflux as measured by simultaneous pharyngeal and distal esophageal 24-hour pH/impedance monitoring on PPI.

As expected and in contrast to pharyngeal reflux, distal GER was strongly inhibited on PPI treatment. The number of acidic reflux episodes was reduced from 26 [17–39] to 3 [1–9] (P < 0.01, paired t-test) and AET was reduced from 2.5 ± 0.5% to 0.8 ± 0.3% (P < 0.01, paired t-test). Conversely, as predicted, the number of weakly acidic reflux episodes increased from 14 [11–21] to 25 [22–31] (P < 0.01, paired t-test). These data show that PPI treatment was effective to reduce GER in these patients.

It is possible that in some patients who were less responsive to the PPI treatment, the symptoms attributed to LPR were not solely due to LPR but due to some other unrelated causes. It follows that the inclusion of such patients may obscure the effect of PPI on pharyngeal reflux. In order to address this concern, we analyzed separately the patients who were the most responsive to the PPI treatment. Because there is unfortunately no universally agreed upon definition of good responders to the PPI therapy in LPR, we took the advantage of the stringent criteria recently published by Yadlapati et al. [8]. These criteria required the normalization of RSI (RSI ≤ 13) and the reduction in RSI by at least 50%. Seven patients who fulfilled these criteria were labeled as good responders. In these good responders, the PPI treatment reduced RSI by 75% from 24 [17–27] to 5 [3–11] (P < 0.01, paired t-test). Individual changes in RSI in good responders are shown in Fig. 3a.

Fig. 3.

The effect of the PPI treatment on symptoms and pharyngeal reflux in patients characterized as good responders vs. non-responders to the PPI treatment. (a) Individual changes in RSI in good responders (**P < 0.01, paired t-test, N = 7). (b) The duration of pharyngeal acid exposure during 24 hours before and after PPI treatment in good responders (two-way repeated measure ANOVA followed by Sidak's multiple comparisons test, *P < 0.05, N = 7) (c) Individual changes in RSI in nonresponders (P > 0.1, paired t-test, N = 7). (d) The duration of pharyngeal acid exposure during 24 hours before and after PPI treatment in nonresponders (two-way repeated measure ANOVA, P = 0.4, N = 7). The duration of pharyngeal acid exposure during 24 hours was calculated as cumulative time during which the pH of pharyngeal reflux was below indicated the pH level. ANOVA, analysis of variance; PPI, proton pump inhibitor; RSI, reflux symptom index.

For comparison, we used a group labeled as nonresponders in who RSI failed to normalize after the PPI treatment (RSI > 13 after the PPI treatment, N = 7). RSI in nonresponders was only insignificantly decreased by 12% by the PPI treatment (30 [26–38] vs 21 [21–34], P > 0.1, paired t-test) (Fig. 3c). Otherwise the group of nonresponders was similar to the group of good responders. RSI before the PPI treatment was not significantly different between the groups (P = 0.2, unpaired t-test). RFS before the PPI treatment in good responders and nonresponders was also similar (10 [9–11] and 11 [8–13], respectively, P = 0.9, unpaired t-test). The effect of PPI on distal GER was also similar between good responders and nonresponders. The number of acidic distal GER episodes before PPI and on PPI was 15 [13–23] vs 1 [0–2] (P < 0.01, paired t-test) in good responders and 16 [11–23] vs 0 [0–7] (P < 0.05, paired t-test) in nonresponders. These results indicate that the difference between good responders and nonresponders was not due to difference in baseline symptoms or laryngeal injury attributed to LPR, or due to insufficient inhibition of GER by PPI (i.e., that acid inhibition was effective in nonresponders). The remaining four patients in who PPI treatment normalized RSI, but the reduction in RSI was less than 50% were denoted as partial responders.

Despite dramatic improvement of the symptoms (RSI) in good responders, we observed no reduction in acidic pharyngeal reflux (Fig. 3a and b). As shown in Fig. 3b, there was no reduction in the duration of pharyngeal acid exposure at any level of pH (the reflux with pH < 6.0 was even increased). Similarly, the number of acidic pharyngeal episodes was not reduced (data not shown). For comparison, the effect of PPI treatment on symptoms (RSI) and pharyngeal reflux in nonresponders is shown in Fig. 3c and d. Again, there was no reduction in the duration (Fig. 3d) or the number of pharyngeal episodes (not shown) in nonresponders.

We next addressed the hypothesis that good responders have more acidic pharyngeal reflux compared to nonresponders before the PPI treatment. We reasoned that if the symptoms respond to PPI, then they are most probable caused by acidic GER, while acidic GER is less important in symptoms which are not responsive to PPI. However, the duration of pharyngeal reflux or the number of pharyngeal reflux episodes at any pH level before PPI treatment were not higher in good responders compared to nonresponders (Fig. 4). Paradoxically, the duration of pharyngeal reflux at pH <6.0 trended to be higher (Fig. 4a) and the number of reflux episodes with pH <6.0 was higher (Fig. 4b) in nonresponders. There was also no difference in the number of impedance-defined reflux episodes before the PPI treatment between good responders and nonresponders. The number of pharyngeal reflux episodes as defined by impedance-based criteria developed by Zerbib et al. [18] before and after the PPI treatment was 2 [1–5] vs 1 [0–2] (P = 0.3, unpaired t-test) in good responders, and 1 [0–1] vs 1 [0–2] (P = 0.5, unpaired t-test) in nonresponders. Similar results were obtained when impedance-based criteria developed by Kawamura et al. [26] were used (data not shown). Thus, the pharyngeal reflux burden before the PPI treatment was not higher in good responders compared to nonresponders.

Fig. 4.

Pharyngeal reflux before the PPI treatment is not higher in patients characterized as good responders compared to nonresponders to the PPI treatment. (a) The duration of pharyngeal acid exposure during 24 hours before the PPI treatment in good responders (N = 7) and nonresponders (N = 7) (two-way repeated measure ANOVA, P = 0.1). (b) The number of pharyngeal acidic reflux episodes with indicated acidity during 24 hours before PPI treatment in good responders and nonresponders. Paradoxically, the number of reflux episodes with pH < 6.0 was higher (two-way repeated measure ANOVA followed by Sidak's multiple comparisons test, **P < 0.01, unpaired t-test) in nonresponders. ANOVA, analysis of variance; PPI, proton pump inhibitor.

Discussion

We found that although the PPI treatment substantially improved the symptoms and laryngeal injury attributed to LPR, it did not affect acidic pharyngeal reflux as measured by simultaneous pharyngeal and esophageal 24-hour pH/impedance monitoring in patients diagnosed with LPR. This was true despite a comprehensive analysis of pharyngeal reflux at various pH levels (pH < 4.0, pH < 4.5, pH < 5.0, pH < 5.5 and pH < 6.0) and impedance-based analysis. Furthermore, separate analysis of a subgroup of patients who experienced a large improvement in symptoms (good responders, RSI reduced by 75%) also did not reveal any effect of the PPI treatment on the pharyngeal reflux. Finally, we also found that the patients who responded to the PPI treatment did not have more pharyngeal reflux than those patients who failed to respond.

We assert that our conclusions are strengthened by the use of an unbiased approach to quantify pharyngeal acid reflux which we developed in our previous study [15]. We did not make a priori assumptions about what pH of pharyngeal reflux reflects LPR causing symptoms and laryngeal injury. Instead, we analyzed pharyngeal reflux events for a complete range of possible pharyngeal pH from pH <6.0 to pH <4.0 (pH < 3.0 was not detected in the pharynx in our studies). For each pH level, we evaluated both the number of pharyngeal reflux episodes and the cumulative duration of pharyngeal exposure to acid of that pH level. Thus, our approach adds additional information to that derived from previous studies in which typically a predetermined pH threshold was considered [4,13,27-29]. In addition to comprehensive analysis of pH, the strengths of our study also include simultaneous pharyngeal and distal esophageal 24-hour pH/impedance monitoring, which established GER as the source of acidic pharyngeal reflux episodes, and the enrollment of patients who were off PPI for at least 3 months prior to the study.

The discrepancy between the large effect of PPI treatment on symptoms and the lack of the effect of PPI on pharyngeal reflux can be due to several factors. First, it is possible that the acidic pharyngeal reflux as detected by simultaneous pharyngeal and esophageal 24-hour pH/impedance monitoring is not suitable to detect pathophysiologically relevant aspects of LPR. Second, it is possible that the symptoms attributed to LPR are only partially caused by LPR and that the acid in the esophagus contributes to these symptoms through additional different mechanisms. Finally, it is also possible that the PPI treatment has at least some placebo effect, which may confuse the analysis [30-32].

A consistent finding in LPR patients is that the pharyngeal reflux with a substantial drop in pH is very rare on 24-hour pH/impedance monitoring. For example, prior to PPI treatment, the median number of pharyngeal reflux episodes with the pH = 5.0 or less was 1 [0–2] (median [IQR]) in good responders. Similar observations were made in other studies. These findings may reflect a scenario in which a severe LPR capable of inducing and sustaining the symptoms and laryngeal damage occurs not every day, but substantially more rarely. For example, if a consequential LPR occurs on average only approximately once in 5 days, the probability of capturing it by a snapshot 24-hour monitoring would be only 20%. It follows that if it is improbable to capture such a rare reflux by the 24-hour monitoring, the 24-hour monitoring would not reveal reduction or elimination of such reflux by the PPI treatment. This is one explanation why the acidic pharyngeal reflux would not be adequate to detect LPR.

Another possible explanation is that acidic pharyngeal reflux is not a good surrogate for LPR. Because direct detection of acid in the larynx is technically challenging, the acid (pH) is instead monitored in the pharynx [18,26,33]. The tacit assumption behind this approach is that if acidic reflux from the esophagus reaches the pharynx, it also penetrates into the larynx. While this assumption is certainly reasonable, there may be unknown variables that influence the extent of penetration of acid into the larynx. For example, the extent of acid penetration may depend not only on the acidity (pH) of reflux, but also on its amount (volume). Therefore, the gastroesophageal reflux episodes that cause a similar decrease in pharyngeal pH and have a similar duration may differ in their penetration into the larynx depending on the volume of refluxed acid. If this is the case, it would be difficult to reliably determine LPR from the pharyngeal reflux. Other variables such as laryngeal defense mechanisms may also influence the consequence of LPR.

The possibility that acid in the esophagus contributes to symptoms attributed to LPR by mechanisms other then LPR may further confuse the issue. For example, the acid in the esophagus may contribute to respiratory symptoms such as cough by activating esophageal vagal nerves, which in turn enhance coughing. The studies of others as well as our previous studies are consistent with such mechanism [34,35]. Similar neural mechanisms may contribute to enhanced sensory perceptions from the laryngopharyngeal area that is attributed to LPR including sore throat and the need to clear the throat. It follows that if there is a substantial contribution of refluxed acid in the esophagus (not larynx) to symptoms attributed to LPR then the PPI treatment may improve these symptoms even without a substantial effect on LPR.

Nonetheless, if the LPR episodes causing symptoms are relatively rare, or 24-hour monitoring of the pharyngeal reflux does not adequately capture LPR, or GER contributes to LPR symptoms by mechanisms other than LPR, then the 24-hour monitoring of pharyngeal reflux may not provide meaningful information about LPR. It would also explain why, despite the dramatic 75% improvement in symptoms, the pharyngeal reflux was not affected by the PPI treatment in good responders. It cannot be excluded that other corrosive components (e.g., pepsin, pancreatic enzymes, bile components), which were not measured in this study, are responsible for laryngeal symptoms and injury [29,36]. Nonetheless, we consider it improbable that such components would be reduced in pharyngeal reflux by the PPI treatment, while the pharyngeal reflux itself and its acidity were not reduced. One cannot exclude the possibility of dietary modification on the symptomatic relief. Although not recommended, patients might have changed their dietary habits by themselves with a consequent decrease in the RSI.

In accordance with recent recommendations and consistent with many previous studies we used PPI b.i.d for 12 weeks [3,13,14,37]. It is well recognized that there is a placebo effect of the PPI treatment in patients with suspected LPR [38]. In the placebo controlled studies of the effect of the PPI treatment on LPR symptoms (RSI), the placebo caused on average less than 20% improvement in RSI [3,5,30-32,38]. This is a substantially smaller effect on RSI than the 50% improvement we found in the whole group of patients. Therefore, while the placebo effect is a potential confounding factor, its effect in our study is probably limited. The placebo effect is probably even less important in the group we labeled as good responders (improvement of RSI by at least 50% and the normalization of RSI to ≤13). While there is no generally accepted definition of good responders to PPI in LPR, there is little doubt that the symptoms in good responders in our study were greatly improved by the PPI therapy (75% reduction in RSI from 24 [17–27] to 5 [3–11]) (Fig. 3a). It would be therefore difficult to argue that acidic GER does not play a major role in the pathogenesis of the symptoms in this group of patients.

Only a few studies used 24-hour pharyngeal pH or pH/impedance monitoring to compare pharyngeal reflux off PPI vs on PPI. In their landmark study, Dulery et al. [9] focused on patients in who the symptoms attributed to LPR did not significantly improve by PPI similar to the group of nonresponders in our study. When we applied identical impedance-based stringent criteria as Dulery et al. [9], we also found that the pharyngeal reflux was extremely rare in nonresponders in our study(median number of pharyngeal reflux episodes off PPI was 1). Similarly to their findings, the pharyngeal reflux in nonresponders was not affected by PPI. Thus, our findings are consistent with those of Dulery et al. [9]. In addition, we also demonstrated that even in good responders, the pharyngeal reflux was rare and was not significantly affected by PPI. On the other hand, our findings are different from those of Reichel et al. [4] who found that pharyngeal reflux was improved by PPI in approximately 50% of patients in a group preselected based on an abnormal pretreatment pharyngeal reflux. In addition to patient selection, distinct methodological approach may also contribute to different findings.

We also aimed to gain better insight into the relationship between LPR and the symptoms attributed to LPR by comparing acidic pharyngeal reflux between the good responders and nonresponders to PPI. The symptoms in nonresponders failed to improve (insignificant RSI reduction by 12%) contrasting to robust improvement in good responders (RSI reduction by 75%). As discussed above, if the symptoms respond to the PPI treatment, then acidic GER probably plays a major role in their pathogenesis. In contrast, acidic GER is probably not a major cause of symptoms that were not improved by PPI. Therefore, we hypothesized that good responders have more LPR. However, we found that prior to the PPI treatment, the acidic pharyngeal reflux was not higher in good responders compared to nonresponders (Fig. 4). The lack of the effect of PPI on the LPR symptoms in nonresponders was not due to the failure of PPI to control GER inasmuch as GER was similarly inhibited in both good responders and nonresponders. These results are in agreement with the conclusion that acidic pharyngeal reflux prior to the treatment does not differentiate between good responders and nonresponders. Indeed, the majority of the studies concluded that pharyngeal pH monitoring does not predict the response to PPI treatment. For example, Yadlapati et al. [8] found that the oropharyngeal pH testing did not predict the response of symptoms to PPI therapy. Similarly, Vaezi et al. [5] and Vaezi [10] showed that dual pH monitoring without impedance did not predict the response to therapy. On the other hand one recent retrospective study reported a positive correlation between the change of RSI and the change of Ryan score (in the upright position only) [39].

Taken together, our data provide further evidence arguing against the utility of 24h pharyngeal pH monitoring in the diagnosis and management of patients with suspected LPR most probably because 24-hour pharyngeal monitoring does not quantitatively capture the pathophysiologically relevant cause of the symptoms in these patients. This is probably caused by a combination of factors including that LPR is rare, pharyngeal pH does not adequately measure LPR and multiple mechanisms of acid-induced symptoms attributed to LPR. Our study has several limitations. First, the evaluation of the RFS was performed by only one ENT physician and therefore we did not determine neither interindividual variability nor the reliability of RFS in our study. Second, the reliability of RSI has also been also questioned and currently new clinical instruments are being developed for the evaluation of LPR symptoms [40]. Finally, as many as relatively high number of patients was lost to follow-up.