Predicting Response to Proton Pump Inhibitors in GERD
Predicting Response to Proton Pump Inhibitors in GERD
The most striking observation in the present study is that no reflux pattern on 24 h pH-impedance monitoring performed off therapy is predictive of response to PPI therapy in GORD. In contrast, it was observed that PPI failure is mainly associated with BMI ≤25 kg/m and presence of functional digestive disorders, even when pathological GOR can be demonstrated by 24 h pH-impedance monitoring.
This study was performed in a population of patients that can be considered to be representative of patients presenting with typical GORD symptoms: low proportion of patients with reflux oesophagitis (14%), high prevalence of overweight (36%) and associated functional disorders (48% with FD and 36% with IBS). Most responders (77%) to PPI were receiving a daily single dose, while the majority of patients (62%) refractory to therapy were receiving a double dose of PPIs, which is very common in France. There were only 35% of patients with abnormal 24 h oesophageal acid exposure, but as a whole, 67% of patients had documented abnormal GOR defined by either abnormal oesophageal acid exposure or positive symptom–reflux association analysis. Therefore one could argue that this population did not present with severe GORD, but with either NERD or hypersensitive oesophagus. Although there are no specific studies to support this assumption, we believe that it is nowadays the most common presentation of patients with GORD, at least those referred to specialised centres for oesophageal investigations. Finally, since it is not easy to justify 24 h pH-impedance monitoring in patients faring well on therapy, we included patients referred for surgery (as preoperative work-up) and patients referred for refractory symptoms who appeared to be adequately controlled after treatment optimisation. We also took the opportunity to include responders in whom the pH-impedance monitoring was performed at baseline of a pharmacological clinical trial. Finally, we performed the analysis in different subgroups of patients. Apart from the whole population, we studied separately the subgroup of patients who reported symptoms during the 24 h pH-impedance monitoring in order to analyse more adequately the role of symptom–reflux association analysis and the subgroup of patients with documented pathological GORD during 24 h pH-impedance monitoring.
The present study is the first outcome study in which the patterns of GOR were determined by 24 h pH-impedance monitoring off PPI. Compared with conventional pH monitoring, impedance provides data on chemical composition (acid, non-acid) as well as the proximal extent of the refluxate within the oesophagus. However, despite the use of this technology, the multivariate analysis taking into account both clinical and physiological parameters could not identify any reflux pattern associated with PPI failure. As an example, reflux events with high proximal extent have been shown to be associated with persisting symptoms in patients receiving PPIs, but the present study failed to demonstrate that patients with a high rate of proximal reflux events were less likely to respond to PPI therapy. Unexpectedly, we could not demonstrate any predictive value of symptom–reflux association analysis with regard to response to PPI therapy, whereas, in the literature, both SI and SAP have been shown to be associated with favourable outcome. However, conflicting data have been reported. Broeders et al have reported similar outcome after fundoplication among patients with a positive and negative SAP. In a study assessing the performance of these indices compared with the omeprazole test, SI was shown to be unrelated to the results of the test, whereas SAP had a positive predictive value of 79% and a specificity of 73%; SI and SAP were therefore considered by the authors themselves to be 'suboptimal' predictors of response to high-dose omeprazole. Finally, a very recent study showed that SI and SAP may be overinterpreted in patients with refractory GORD, especially when low reflux rates are observed at 24 h pH-alone or pH-impedance monitoring. Taken together, these results raise concerns about the validity and/or usefulness of the indices, at least to predict the response to both medical and surgical therapy. However, we do believe that there is still room for pH-impedance monitoring in patients with refractory GORD on PPI therapy; these indices should be considered as complementary to the 'quantitative' evaluation of reflux, since they can help to identify more patients with symptoms possibly related to GORD.
When the whole population was considered, although the univariate analysis showed that many factors were associated with PPI failure, the multivariate analysis showed that only normal or low BMI, FD and the absence of oesophagitis at endoscopy were significantly and independently associated with PPI failure. Although it is now well established that absence of oesophagitis is associated with poorer response to PPI, the relationship between BMI and response to PPIs was initially unexpected, but has been suggested by two recent studies. In a study by Fletcher et al in 105 patients with upper gastrointestinal tract (GI) symptoms and normal endoscopy, BMI >25 kg/m and LOS pressure were the only independent factors associated with response to PPI therapy. BMI had a similar predictive value to either 24 h oesophageal pH monitoring or manometry. Unlike the present study, Fletcher et al included patients with upper GI symptoms, not only heartburn and regurgitation, but it was observed that predominant symptom and symptom subgroups were unhelpful in predicting the response to PPIs. Heading et al performed an open study of 8 weeks pantoprazole 40 mg daily in 1888 patients presenting with symptoms 'considered by the investigating physician to justify a diagnosis of GOR' and also observed poorer treatment responses to be independently associated with lower BMI. Our study is therefore the third to report similar results, whatever the subgroup of patients considered. The mechanisms by which BMI may influence the response to PPIs remain to be elucidated. Hence, high BMI has been clearly associated with the development of reflux symptoms and complications through different putative mechanisms such as increased transient LOS relaxations rate, gastro-oesophageal pressure gradient and oesophago–gastric junction disruption—that is, separation of LOS and crural diaphragm leading to hiatal hernia. Therefore high BMI may increase the probability that symptoms are related to GORD, but it is of note that BMI ≤25 kg/m was also independently associated with PPI failure in patients with demonstrated GORD on pH-impedance monitoring.
Several studies have reported the negative impact of functional gastrointestinal disorders on treatment of reflux symptoms. We have observed that FD and IBS were strongly associated with PPI failure, even in patients with documented abnormal GOR with pH-impedance monitoring. The prevalence of dyspeptic symptoms is high in patients with functional heartburn—that is, with normal oesophageal acid exposure and negative symptom–reflux association analysis. However, in the present study, the role of FD as a predictor of PPI failure was also demonstrated in patients with documented GORD. Although we paid close attention to the nature of the symptoms that persisted during therapy by means of a standardised questionnaire, some patients may have dyspeptic symptoms that may be misinterpreted as reflux symptoms. A recent study that assessed dyspeptic symptoms in patients treated by PPIs did not report any effect of ulcer-like or dysmotility-like symptoms on treatment outcome in patients with upper GI symptoms. The reasons for these discrepancies are not clear, but may be related to different diagnostic criteria for dyspepsia and to the fact that we included patients with typical GOR symptoms as the primary complaint. We also observed that IBS was a predictor of PPI failure in the subgroup of patients with documented abnormal GOR. Similar results have been recently reported by Heading et al in patients included on the basis of symptoms only; another study demonstrated that IBS predicted worse quality of life after PPI therapy. The interpretation of this finding is far from easy, since we can reasonably rule out the possibility of symptom misinterpretation. It may be hypothesised that persistent reflux symptoms and IBS symptoms share the same underlying mechanisms (eg, increased visceral perception), since the two conditions often coexist. Finally, we believe that the presence of functional GI disorders in patients with refractory GORD should be taken into account in future trials with reflux inhibitors.
Our study has several limitations. First, the absence of standardisation of PPI therapy and the use of a recall questionnaire may represent a bias for interpretation of our data. However, regarding the modalities of inclusion in the study together with the use of a standardised questionnaire, the probability of misinterpretation of symptoms is relatively low. Second, we did not assess the level of anxiety in our patients. Anxiety, usually assessed by means of the Hospital Anxiety and Depression Scale, has been shown to be associated with PPI failure. However, conflicting results have been recently reported, and the level of anxiety was normal in a population of patients with refractory GORD included in a pharmacological study. Moreover, the possibility that high anxiety scores may be induced by the persisting reflux symptoms themselves cannot be ruled out.
In conclusion, the results of 24 h pH-impedance monitoring performed off therapy are not predictive of response to PPIs. In patients presenting with typical GORD symptoms as predominant complaint, BMI ≤25 kg/m, absence of oesophagitis and functional digestive disorders are the main factors associated with PPI failure through mechanisms that remain to be further elucidated.
Discussion
The most striking observation in the present study is that no reflux pattern on 24 h pH-impedance monitoring performed off therapy is predictive of response to PPI therapy in GORD. In contrast, it was observed that PPI failure is mainly associated with BMI ≤25 kg/m and presence of functional digestive disorders, even when pathological GOR can be demonstrated by 24 h pH-impedance monitoring.
This study was performed in a population of patients that can be considered to be representative of patients presenting with typical GORD symptoms: low proportion of patients with reflux oesophagitis (14%), high prevalence of overweight (36%) and associated functional disorders (48% with FD and 36% with IBS). Most responders (77%) to PPI were receiving a daily single dose, while the majority of patients (62%) refractory to therapy were receiving a double dose of PPIs, which is very common in France. There were only 35% of patients with abnormal 24 h oesophageal acid exposure, but as a whole, 67% of patients had documented abnormal GOR defined by either abnormal oesophageal acid exposure or positive symptom–reflux association analysis. Therefore one could argue that this population did not present with severe GORD, but with either NERD or hypersensitive oesophagus. Although there are no specific studies to support this assumption, we believe that it is nowadays the most common presentation of patients with GORD, at least those referred to specialised centres for oesophageal investigations. Finally, since it is not easy to justify 24 h pH-impedance monitoring in patients faring well on therapy, we included patients referred for surgery (as preoperative work-up) and patients referred for refractory symptoms who appeared to be adequately controlled after treatment optimisation. We also took the opportunity to include responders in whom the pH-impedance monitoring was performed at baseline of a pharmacological clinical trial. Finally, we performed the analysis in different subgroups of patients. Apart from the whole population, we studied separately the subgroup of patients who reported symptoms during the 24 h pH-impedance monitoring in order to analyse more adequately the role of symptom–reflux association analysis and the subgroup of patients with documented pathological GORD during 24 h pH-impedance monitoring.
The present study is the first outcome study in which the patterns of GOR were determined by 24 h pH-impedance monitoring off PPI. Compared with conventional pH monitoring, impedance provides data on chemical composition (acid, non-acid) as well as the proximal extent of the refluxate within the oesophagus. However, despite the use of this technology, the multivariate analysis taking into account both clinical and physiological parameters could not identify any reflux pattern associated with PPI failure. As an example, reflux events with high proximal extent have been shown to be associated with persisting symptoms in patients receiving PPIs, but the present study failed to demonstrate that patients with a high rate of proximal reflux events were less likely to respond to PPI therapy. Unexpectedly, we could not demonstrate any predictive value of symptom–reflux association analysis with regard to response to PPI therapy, whereas, in the literature, both SI and SAP have been shown to be associated with favourable outcome. However, conflicting data have been reported. Broeders et al have reported similar outcome after fundoplication among patients with a positive and negative SAP. In a study assessing the performance of these indices compared with the omeprazole test, SI was shown to be unrelated to the results of the test, whereas SAP had a positive predictive value of 79% and a specificity of 73%; SI and SAP were therefore considered by the authors themselves to be 'suboptimal' predictors of response to high-dose omeprazole. Finally, a very recent study showed that SI and SAP may be overinterpreted in patients with refractory GORD, especially when low reflux rates are observed at 24 h pH-alone or pH-impedance monitoring. Taken together, these results raise concerns about the validity and/or usefulness of the indices, at least to predict the response to both medical and surgical therapy. However, we do believe that there is still room for pH-impedance monitoring in patients with refractory GORD on PPI therapy; these indices should be considered as complementary to the 'quantitative' evaluation of reflux, since they can help to identify more patients with symptoms possibly related to GORD.
When the whole population was considered, although the univariate analysis showed that many factors were associated with PPI failure, the multivariate analysis showed that only normal or low BMI, FD and the absence of oesophagitis at endoscopy were significantly and independently associated with PPI failure. Although it is now well established that absence of oesophagitis is associated with poorer response to PPI, the relationship between BMI and response to PPIs was initially unexpected, but has been suggested by two recent studies. In a study by Fletcher et al in 105 patients with upper gastrointestinal tract (GI) symptoms and normal endoscopy, BMI >25 kg/m and LOS pressure were the only independent factors associated with response to PPI therapy. BMI had a similar predictive value to either 24 h oesophageal pH monitoring or manometry. Unlike the present study, Fletcher et al included patients with upper GI symptoms, not only heartburn and regurgitation, but it was observed that predominant symptom and symptom subgroups were unhelpful in predicting the response to PPIs. Heading et al performed an open study of 8 weeks pantoprazole 40 mg daily in 1888 patients presenting with symptoms 'considered by the investigating physician to justify a diagnosis of GOR' and also observed poorer treatment responses to be independently associated with lower BMI. Our study is therefore the third to report similar results, whatever the subgroup of patients considered. The mechanisms by which BMI may influence the response to PPIs remain to be elucidated. Hence, high BMI has been clearly associated with the development of reflux symptoms and complications through different putative mechanisms such as increased transient LOS relaxations rate, gastro-oesophageal pressure gradient and oesophago–gastric junction disruption—that is, separation of LOS and crural diaphragm leading to hiatal hernia. Therefore high BMI may increase the probability that symptoms are related to GORD, but it is of note that BMI ≤25 kg/m was also independently associated with PPI failure in patients with demonstrated GORD on pH-impedance monitoring.
Several studies have reported the negative impact of functional gastrointestinal disorders on treatment of reflux symptoms. We have observed that FD and IBS were strongly associated with PPI failure, even in patients with documented abnormal GOR with pH-impedance monitoring. The prevalence of dyspeptic symptoms is high in patients with functional heartburn—that is, with normal oesophageal acid exposure and negative symptom–reflux association analysis. However, in the present study, the role of FD as a predictor of PPI failure was also demonstrated in patients with documented GORD. Although we paid close attention to the nature of the symptoms that persisted during therapy by means of a standardised questionnaire, some patients may have dyspeptic symptoms that may be misinterpreted as reflux symptoms. A recent study that assessed dyspeptic symptoms in patients treated by PPIs did not report any effect of ulcer-like or dysmotility-like symptoms on treatment outcome in patients with upper GI symptoms. The reasons for these discrepancies are not clear, but may be related to different diagnostic criteria for dyspepsia and to the fact that we included patients with typical GOR symptoms as the primary complaint. We also observed that IBS was a predictor of PPI failure in the subgroup of patients with documented abnormal GOR. Similar results have been recently reported by Heading et al in patients included on the basis of symptoms only; another study demonstrated that IBS predicted worse quality of life after PPI therapy. The interpretation of this finding is far from easy, since we can reasonably rule out the possibility of symptom misinterpretation. It may be hypothesised that persistent reflux symptoms and IBS symptoms share the same underlying mechanisms (eg, increased visceral perception), since the two conditions often coexist. Finally, we believe that the presence of functional GI disorders in patients with refractory GORD should be taken into account in future trials with reflux inhibitors.
Our study has several limitations. First, the absence of standardisation of PPI therapy and the use of a recall questionnaire may represent a bias for interpretation of our data. However, regarding the modalities of inclusion in the study together with the use of a standardised questionnaire, the probability of misinterpretation of symptoms is relatively low. Second, we did not assess the level of anxiety in our patients. Anxiety, usually assessed by means of the Hospital Anxiety and Depression Scale, has been shown to be associated with PPI failure. However, conflicting results have been recently reported, and the level of anxiety was normal in a population of patients with refractory GORD included in a pharmacological study. Moreover, the possibility that high anxiety scores may be induced by the persisting reflux symptoms themselves cannot be ruled out.
In conclusion, the results of 24 h pH-impedance monitoring performed off therapy are not predictive of response to PPIs. In patients presenting with typical GORD symptoms as predominant complaint, BMI ≤25 kg/m, absence of oesophagitis and functional digestive disorders are the main factors associated with PPI failure through mechanisms that remain to be further elucidated.