The impact of pre-endoscopy proton pump inhibitor use on the classification of non-erosive reflux disease and erosive oesophagitis

Authors

  • S. Gaddam,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    Search for more papers by this author
  • S. Wani,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • H. Ahmed,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • P. Maddur,

    1. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • S. B. Hall,

    1. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • N. Gupta,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • S. R. Puli,

    1. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • A. Higbee,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    Search for more papers by this author
  • A. Rastogi,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • A. Bansal,

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author
  • P. Sharma

    1. Division of Gastroenterology and Hepatology, Veterans Affairs Medical Center, Kansas City, MO, USA.
    2. Division of Gastroenterology and Hepatology, University of Kansas School of Medicine, Kansas City, MO, USA.
    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Corrigenda Volume 34, Issue 11-12, 1356, Article first published online: 16 November 2011

Dr P. Sharma, Division of Gastroenterology and Hepatology, Department of Veterans Affairs Medical Center, 4801 E. Linwood Blvd, Kansas City, MO 64128-2295, USA.
E-mail: psharma@kumc.edu

Abstract

Aliment Pharmacol Ther 2010; 32: 1266–1274

Summary

Background  Factors associated with non-erosive reflux disease (NERD) and erosive oesophagitis (EO) are incompletely understood and the overlap between the two entities is debated.

Aim  To compare clinical, demographic, and endoscopic findings in a large cohort of NERD and EO patients.

Methods  After they completed a validated GERD questionnaire, patients who presented for index endoscopy were enrolled and categorized as NERD or EO. Analysis was performed using Chi-square, Mann–Whitney U-test and multivariate logistic regression.

Results  A total of 696 GERD patients [455 (65.4%) NERD; 241 (34.6%) EO]; mean age 57 years; 92% men and 82% Caucasian were prospectively enrolled. Using logistic regression, patients on PPI were more likely to be classified as NERD (OR: 3.2; < 0.001). NERD patients were older (OR: 1.50; = 0.05), less likely to have nocturnal symptoms (OR: 0.63; = 0.04) and hiatal hernia (OR: 0.32; < 0.001). Compared with PPI-naïve NERD patients, those on PPI were more likely to have nocturnal symptoms (69% vs. 29%, = 0.048) and less likely to have mild-moderate symptoms (63% vs. 79%, < 0.001) – similar to the EO group.

Conclusions  Pre-endoscopy PPI usage contributes significantly to the classification of GERD patients into the NERD-phenotype. NERD patients on PPI therapy demonstrate some features that are significantly different from PPI-naïve patients, but similar to EO patients. This observation supports the notion that some PPI NERD patients are actually healed EO patients, and that an overlap does exist between the GERD phenotypes.

Introduction

Gastro-oesophageal reflux disease (GERD) is a common disorder in the Western countries that develops when the reflux of stomach contents causes troublesome symptoms and/or complications. It is associated with a huge economic burden and decreased quality of life.1–5 In a recent systematic review, the prevalence of GERD in western countries was estimated to be 20% when GERD was defined as symptoms of heartburn and/or regurgitation occurring at least weekly with an incidence of approximately 5 per 1000 person years.3 GERD is the most common GI disorder accounting for 10 billion dollars per year in the United States as direct costs,1 whereas indirect costs are estimated to be as much as 75 billion dollars per year.2 In addition, GERD is associated with worrisome consequences such as strictures, Barrett’s oesophagus (BO) and oesophageal adenocarcinoma.6

Based on the findings at endoscopy, GERD patients are generally categorized into one of the following phenotypes: non-erosive reflux disease (NERD) or erosive oesophagitis (EO).7 It is estimated that 40–50% of patients with typical reflux symptoms in tertiary centres and only 20–30% in primary care practice have erosive oesophagitis.8 Patients classified as NERD have similar symptom burden as EO and an impact on quality of life that is at least comparable to that of EO.9 However, patients with NERD have a much lower response to proton pump inhibitor (PPI) therapy.10 They are also more likely to need further evaluation of GERD symptoms, which includes evaluation for functional heartburn, non-acid reflux,11 gastric hypomotility12 or oesophageal hypersensitivity.13

Clinical predictors for presence of NERD are not well defined.14 Data are also limited on why some patients with reflux show no endoscopic signs of oesophagitis and what factors influence severity of the disease. Although some risk factors for erosive oesophagitis such as gender, smoking, alcohol consumption and obesity have been evaluated, they are not well understood and endoscopic examination remains essential for an accurate assessment and classification of GERD.

Finally, it is common practice to treat GERD patients with PPIs, in addition to lifestyle modification at the time of their visit with their primary care physician. Most patients who are refractory to PPI therapy are then referred to a gastroenterologist for further evaluation. It is possible that this PPI use prior to endoscopy may result in healed erosive oesophagitis and may misclassify EO patients to have the NERD phenotype.

The aims of this study were to evaluate prospectively patients with GERD using a validated GERD questionnaire and endoscopic examination and to compare clinical, demographic and endoscopic findings among patients with NERD and EO, to evaluate if PPI usage was responsible for higher prevalence of NERD compared to EO, and to establish the predictors for NERD and EO in those patients not on PPI therapy at the time of upper endoscopy.

Methods

Patients

This study was conducted at the Veterans Affairs Medical Center, Kansas City, Missouri. A total of 926 consecutive patients presenting for their first ever endoscopic evaluation of reflux symptoms between October 2003 and June 2009 were recruited. Patients referred for EGD and those who had ≥6 months of GERD symptoms were prospectively enrolled in the study. Eligible patients were asked to complete a validated GERD questionnaire.15 Patients were classified to have GERD based on the response on the validated GERD questionnaire – GERQ.15 Patients with one or more of the following criteria were excluded from this study: prior upper endoscopic evaluation, inability to provide written informed consent, evidence of oesophageal varices, severe oesophageal strictures and prior history of oesophageal or gastric surgery. The study was approved by the Human Subjects Committee.

Questionnaire

All patients completed a validated GERD questionnaire (GERQ).15 GERQ is designed as a self-report instrument to measure detailed symptoms of heartburn and acid regurgitation experienced over the previous years and to collect pertinent medical history. It is easy to complete and in previous testing, this instrument was demonstrated to be reliable, with a median κ statistic for symptom items of 0.71 (interquartile range: 0.63–0.81). This questionnaire identifies the onset for GERD symptoms and grades the frequency and severity of the symptoms of heartburn and acid regurgitation. Additional information regarding age, gender, race, weight and height, body mass index (BMI), health habits such as tobacco smoking, medication use such as proton pump inhibitors (PPIs), H2-receptor blockers, aspirin (ASA) and nonsteroidal anti-inflammatory drugs (NSAIDs) and family history of heartburn or oesophageal disease were also collected.

Endoscopy

Upper endoscopy was performed using standard video endoscopes (Olympus GIF-160, GIF – 180; Olympus America Inc, Center Valley, PA, USA). The distal oesophagus was carefully inspected for the presence of mucosal breaks or columnar lined oesophagus. Diagnosis of erosive oesophagitis was based on the endoscopic appearance of definite mucosal single or confluent breaks. EO was graded according to the Los Angeles classification.16 NERD was defined using the Genval Workshop definition as patients with GERD symptoms without either Barrett’s oesophagus or oesophageal mucosal breaks or erosions on endoscopy.17 The gastro-oesophageal junction was identified by the proximal margin of the gastric folds coinciding with the pinch at the end of the tubular oesophagus. The presence of hiatal hernia (HH) was noted and size was measured from the gastro-oesophageal junction to the diaphragmatic pinch.

Definitions and measurement

The following definitions were used to identify symptoms: (i) heartburn, a burning pain or discomfort behind the breast bone in the chest; (ii) acid regurgitation, a bitter or sour tasting fluid coming into the throat or mouth. Nocturnal reflux was assessed by questions that include either (i) awakening at night in the last year due to heartburn or (ii) awakening at night due to acid regurgitation within the last year. Symptom frequency was measured on the following scale: (1) none in past year, (2) less than once a month, (3) about once a month, (4) about once a week, (5) several times a week and (6) daily. Symptom severity was assessed on a 4-point scale as follows: mild (can be ignored), moderate (cannot be ignored, but does not affect lifestyle), severe (affects lifestyle) and very severe (markedly affects lifestyle). Symptom duration was measured as: (1) in the last 6 months, (2) 7 months to 1 year ago, (3) more than 1 year to 2 years ago, (4) more than 2 years to 5 years ago, (5) more than 5 years to 10 years ago, (6) more than 10 years to 20 years ago and (7) more than 20 years ago.

Statistical analyses

Demographic, clinical and endoscopic findings were entered into a computerized database. A statistical software program (spss, Version 15.0; SPSS Inc., Chicago, IL, USA) was used for data management and analysis. Chi-square test was employed to study the effect and possible correlation among race, gender, smoking, symptom severity and frequency, presence of nocturnal symptoms, use of PPIs and ASA/NSAIDs, presence of hiatal hernia in patients with NERD and EO. Continuous variables were evaluated by using the Mann–Whitney U-test. All variables that attained a P value of <0.1 were then analysed using logistic regression to evaluate for independent predictors of NERD and EO. Odds ratios and their 95% confidence intervals (CI) were calculated for each estimate. Statistical significance was defined as < 0.05.

To find predictors of NERD and to compare NERD with EO in a PPI-naïve population, a subgroup analysis was performed in patients who were not on PPI at the time of endoscopy. Finally, we also compared NERD patients who were PPI-naïve with those on PPI therapy.

Results

Nine hundred and twenty-six patients were enrolled in the study after completion of the GERD questionnaire. Of the 926 patients, three were excluded for missing relevant information. Two hundred and twenty-seven (24.5%) were excluded for columnar lined epithelium on endoscopy and/or the confirmed presence of intestinal metaplasia on biopsy. A total of 696 patients were included in the final analysis (Figure 1).

Figure 1.

 Flowchart of study patients.

Patient characteristics

Of a total of 696 patients, 455 (65.4%) presented with NERD, while 241 (34.6%) presented with EO. Mean age (range) was 56.8 (24–93) years and the mean BMI was 29.72 (14.8–53.4). Most of the subjects were men (620, 92.3%). Five hundred and fifty described their race as Caucasian (81.8%), and 105 as African-American (15.6%). Of these patients, 91.6% had heartburn and 76.4% experienced regurgitation. Of the entire study group, 439 (68.1%) suffered from nocturnal reflux symptoms. Three hundred and one patients (43.8%) were found to have hiatal hernia on upper endoscopy. Baseline demographic and clinical characteristics including endoscopic findings are summarized in Table 1.

Table 1.   Baseline demographics, clinical and endoscopic characteristics of the study population
Variablen (%)
Total number696
Mean Age (range)56.87 (24–93)
Male (n, %)620 (92.3%)
Race
 Caucasians (n, %)550 (81.8%)
 African-Americans105 (15.6%)
 Others17 (2.5%)
Mean BMI (range)29.72 (14.75–53.42)
Smoking (past or present) (n, %)449 (70.7%)
Patients on PPI (n, %)272 (45.0%)
Patients on ASA/NSAID (n, %)387 (59.0%)
Number of patients with NERD (n, %)455 (65.4%)
Number of patients with heartburn (n, %)591 (91.6%)
Number of patients with acid regurgitation (n, %)492 (76.4%)
Number of patients with both heartburn and regurgitation (n, %)452 (70.7%)
Number of patients with nocturnal reflux (n, %)439 (68.1%)
Number of patients with hiatal hernia (n, %)301 (43.8%)

Comparison of demographic, clinical and endoscopic findings in NERD and EO patients

Four hundred and fifty-five patients with NERD and 241 patients with EO were compared. Table 2a shows the univariate comparisons of patients with EO and NERD. On univariate analysis, NERD patients were significantly more likely to be female gender (95.2% vs. 90.7%, = 0.047), older (41.8% vs. 32%, < 0.011), have less frequent symptoms, (>1 per week) (68.7% vs. 78.5%; = 0.01) and less nocturnal symptoms (64.4% vs. 75%; = 0.006). Patients with NERD were more likely to be on PPIs compared with EO patients (53.3% vs. 29.3%; < 0.001). Also, patients with NERD were on PPI for a longer mean (s.d.) duration when compared with EO [7.2 (17.6) vs.3 (9.4) months (< 0.001)]. In addition, hiatal hernia was less frequent in patients with NERD than in those with EO (36.1% vs. 58.5%, = <0.001). No significant difference was identified between the two groups of patients with regard to race, BMI, use of ASA/NSAIDs or smoking.

Table 2.    (a) Variables associated with non-erosive oesophagitis: results of univariate analysis. (b) Logistic regression model: Independent predictors of NERD
(a)
VariableReferenceNERDEOOR (95% of CI)P-value
Mean age (s.d.) years 57.72 (12.72)54.95 (13.42) 0.007
Age more than 60 yearsLess than 60 years189 (41.8%)77 (32%)1.53 (1.10–2.13)0.011
Male (%)Female402 (90.7%)218 (95.2%)0.50 (0.25–0.98)0.047
Caucasians (n, %)Non-Caucasians357 (80.8%)193 (83.9%)0.81 (0.53–1.23)0.344
Mean BMI (s.d.) 29.66 (5.4)29.78 (5.5) 0.952
BMI >40BMI ≤4016 (3.6%)14 (6.1%)0.58 (0.28–1.21)0.144
Current or past smokers (n, %)Nonsmokers305 (72.1%)144 (67.9%)1.22 (0.85–1.75)0.28
Number on PPI therapy (n, %)Not on PPI211 (53.3%)61 (29.3%)2.75 (1.92–3.93)<0.001
ASA/NSAID therapy (n, %)Not on ASA/NSAID256 (59.5%)131 (58%)1.07 (0.769–1.479)0.698
Symptom duration < 1 yearOther duration67 (16%)36 (16.1%).99 (.64–1.5)0.98
Symptom duration 1–5 yearsOther duration204 (48.7%)118 (52.7%).85 (.62–1.2)0.34
Symptom duration >5 yearsOther duration148 (35.3%)70 (31.3%)1.2 (.85–1.70)0.3
Symptom severity: Mild-to-moderateSevere-to-very severe301 (72%)144 (64.6%)1.41 (0.99–1.99)0.06
Symptom frequency >1 per week≤1 per week287 (68.7%)175 (78.5%).60 (.41–.88)0.01
Nocturnal symptoms (n, %)No nocturnal symptoms271 (64.4%)168 (75%)0.60 (0.42–0.87)0.006
Hiatal hernia present (n, %)No hiatal hernia163 (36.1%)138 (58.5%)0.40 (0.29–0.56)<0.001
(b)
VariableOR95% CIP-value
PPI usage3.192.14–4.77<0.001
Nocturnal symptoms0.630.41–0.970.035
Age more than 60 years1.491.01–2.240.049
Hiatal hernia0.320.22–0.48<0.001

On multivariate logistic regression analysis, the independent predictors of NERD were PPI usage prior to endoscopy (OR: 3.19, 95% CI: 2.14–4.77, < 0.001), absence of nocturnal symptoms (OR: 0.63; 95% CI: 0.41–0.97, = 0.035), age 60 years or older (OR: 1.50; 95%CI: 1.00–2.24) and absence of hiatal hernia (OR: 0.324; 95% CI: 0.22–0.48) (Table 2b).

Subgroup analysis in PPI-naïve patients

As PPI affected the prevalence of NERD in our analysis, we performed a subgroup analysis by excluding all the patients taking PPI at the time of endoscopy. A data analysis was performed on 332 patients. Of 332 patients, 184 (55.4%) were classified into the NERD phenotype (Supporting Information Table S1).

On univariate analysis, patients with NERD were less likely to experience frequent symptoms (more than once a week) (75.4% vs. 85.3%, = 0.034), less likely to have nocturnal symptoms (59.2% vs. 74.3%, = 0.006) and more likely to have symptoms for greater than 5 years (38.7% vs. 27.2%, = 0.035). Also, NERD patients were less likely to have a hiatal hernia (32.4% vs. 60.8%, = <0.001). There were no statistical differences with respect to gender, age, race, BMI, smoking or NSAID/ASA usage between the two groups.

On multivariate logistic regression, predictors for NERD were symptom duration greater than 5 years (OR = 2.07; 95% CI = 1.21–3.53, = 0.008), lack of frequent symptoms (more than once per week) (OR = 0.51; 95% CI 0.27–0.98, = 0.042) and lack of nocturnal symptoms (OR = 0.49; 95% CI = 0.28–0.84, = 0.01). Also, NERD patients were less likely to have hiatal hernia compared with EO patients (OR = 0.28; 95% CI = 0.17–0.46, = <0.001).

Subgroup analysis comparing NERD patients on and off PPI

Two hundred and eleven NERD patients on PPI therapy were compared with 185 PPI-naïve NERD patients. NERD patients on PPI were more likely to have nocturnal symptoms (69.5% vs. 59.2%, P = 0.048), but less likely to have mild or moderate symptoms (63.1% vs. 79.2%, < 0.001). However, there were no statistical differences between the two groups with respect to gender, age, race, BMI, smoking or NSAID/ASA usage. Results of the comparison are summarized in Table 3.

Table 3.   Differences in patient characteristics between NERD patients on and off PPI therapy
VariableReferenceOn PPI (n = 211)Off PPI (n = 185)P-value
Mean age (s.d.) years 56.6 (12.7)59.1 (12.8)0.075
Age more than 60 yearsLess than 60 years84 (39.8%)85 (45.9%)0.224
Male (%)Female182 (87.9%)165 (93.2%)0.085
Caucasians (n, %)Non-Caucasians171 (82.6%)143 (81.3%)0.79
Current or past smokers (n, %)Nonsmokers149 (77.6%)119 (68.6%)0.059
ASA/NSAID therapy (n, %)Not on ASA/NSAID120 (60.9%)104 (59.4%)0.83
Symptom duration < 1 yearOther duration29 (14.8%)29 (17.3%)0.567
Symptom duration 1–5 yearsOther duration103 (52.6%)74 (44%)0.115
Symptom duration >5 yearsOther duration64 (32.7%)65 (38.7%)0.272
Symptom severity: Mild-to-moderateSevere to very severe123 (63.1%)133 (79.2%)0.001
Symptom frequency >1 per week≤1 per week133 (67.9%)126 (75.4%)0.13
Nocturnal symptoms (n, %)No nocturnal symptoms137 (69.5%)100 (59.2%)0.048
Hiatal hernia present (n, %)No hiatal hernia76 (36.2%)59 (32.4%)0.457

Discussion

Despite NERD being the most common phenotypic presentation of GERD,7 there are limited data on clinical and demographic features associated with these findings. Although there is evidence that patients with NERD have lower response rate and a longer lag time to respond to PPI therapy,10, 18 there are sparse data on the impact of pre-endoscopy usage of PPI on the phenotypic presentation of GERD at endoscopy.

This study found that PPI use prior to endoscopy contributes significantly to classification of GERD patients into the NERD phenotype suggesting that there is an overlap between these entities (Figure 3). NERD patients present with less frequent symptoms, longer duration of symptoms and less nocturnal symptoms. Finally, NERD patients are less likely to have a hiatal hernia when compared with EO (Figure 2).

Figure 3.

 Pie chart represents a conceptual framework of NERD and EO patients. The overlap area represents the EO patients who are diagnosed as NERD at endoscopy in patients who are on PPI therapy (Estimates based on comparing prevalence of NERD and EO in the overall population with PPI-naïve patients). The prevalence and 95% CI are provided.

Figure 2.

 Significant differences between NERD and EO patients on multivariate analysis (all < 0.05).

Our study shows that the usage of PPI prior to endoscopy is an independent risk factor for patients with GERD symptoms to be classified as NERD. Patients with NERD had three times more odds of being on PPI when compared with patients with EO. The two likely explanations are that PPI therapy may have healed EO prior to endoscopy or that difficult symptom control in the NERD group may require PPI therapy. Current practices demonstrate that most GERD patients are empirically treated with PPI upon initial presentation. It is possible that the increased usage of PPIs in patients with reflux symptoms led to healed mucosal lesions by the time of endoscopy and caused the patient to be misclassified as NERD. Previously, it has been suggested that NERD, EO and BO are distinct phenotypes.7, 19, 20 Progression from one type to another is considered a relatively uncommon occurrence. However, our study shows that some of the EO patients may be misclassified as NERD at endoscopy, especially those patients who are on a PPI (Figure 3). Secondly, poor response and persistent reflux symptoms on PPI therapy in the NERD group made them more likely to seek medical attention and hence endoscopic evaluation. Moreover, we found that EO patients who were on a PPI were more likely to have milder (Grades A and B) forms of EO compared with those who are not on a PPI (25.9% vs. 28.6%= N.S.). This is in agreement with our hypothesis that PPI use prior to endoscopy can cause healing of mucosal breaks thus causing patients with milder EO to present as NERD and those with severe EO (Grades C and D) as milder EO.

We also compared clinical, demographic and endoscopic findings between NERD patients on and off PPI. NERD patients on PPI therapy demonstrated some important features that were significantly different from the PPI-naïve NERD patients and similar to those of EO patients. They were more likely to have nocturnal symptoms (69.5% vs. 29.2%, P = 0.048) and less likely to describe their symptoms as mild-to-moderate (63.1% vs. 79.2%, < 0.001), findings that are similar to the EO group. This strengthens the argument that the NERD phenotype at index endoscopy while on PPI therapy includes EO patients with healed mucosal erosions or breaks and that an overlap does exist between the GERD phenotypes. Furthermore, this explains the higher prevalence of NERD in the group on PPI therapy when compared with PPI-naïve population (65% vs. 55%).

Non-erosive reflux disease patients have lower and slower response rates to PPI compared with EO and this has been demonstrated in previous published studies.10, 18 In a systematic review of literature, seven clinical trials evaluating heart burn resolution in NERD were identified. Pooled symptomatic response rates after 4 weeks of PPI therapy was 36.7 (95% CI: 34.1–39.3) in NERD patients and 55.5 (95% CI: 51.5–59.5) in those with EO.10 Dean et al. describe that patients with NERD also demonstrate longer lag-time to sustained symptom response when compared with patients with EO.10 In a more recent study by Lee et al., it was shown that after 8 weeks of PPI therapy, the response rate in NERD patients was 83.3%, whereas in EO patients, it was 94% (= 0.002).18 In a Cochrane review by Moayyedi et al., it was found that PPI therapy is most effective in healing erosive oesophagitis.21 However, there are limited data on the impact of pre-endoscopy PPI on the phenotypic presentation at endoscopy.

It has been previously reported that male gender,9, 22–24 older age,22, 23, 25 obesity,22, 26 smoking18, 22, 23 and alcohol use18, 22 are risk factors for the development of EO. Other associated risk factors include longer duration of symptoms,22 more frequent and severe symptoms of GERD and hiatal hernia.18 Nocturnal reflux has also been associated with severe forms of erosive oesophagitis.27 In studies comparing NERD with EO, NERD was more likely to be associated with younger age, female gender, higher BMI, Helicobacter pylori infection, higher education, absence of history of smoking or alcohol intake and with a shorter GERD history.22, 23, 25 BMI has been shown to be associated with EO.26 Numerous case–control,22, 28–30 cross-sectional31 and cohort studies32 have shown a positive association of obesity with erosive oesophagitis. However, more recent meta-analysis suggests that the risk of GERD symptoms is associated with abdominal obesity (waist circumference) independent of BMI.33 We validated the previously reported association of EO phenotype with frequent symptoms,34 nocturnal symptoms27 and hiatal hernia,18 but failed to show an association with BMI. This can be explained by the fact that we did not include measurement of waist circumference in our study. It is possible that the visceral adiposity is correlated with GERD phenotypes rather than BMI.

The definitions of NERD have varied over the past years. Traditionally, patients with GERD are classified as EO or NERD based on the phenotypic presentation at endoscopy. Genval Workshop (1999) suggested the use of NERD for only those patients who satisfy the definition of GERD without evidence of BO or mucosal breaks at endoscopy.17 Later (2001), other studies have suggested including only those patients with typical symptoms of GERD caused by intra-oesophageal reflux into the NERD category.35 More recently (2009), the Vevey NERD Consensus Group defined NERD as ‘a subcategory of GERD characterized by troublesome reflux-related symptoms in the absence of oesophageal mucosal erosions/breaks at conventional endoscopy and without recent acid-suppressive therapy’.36 Most studies in the past have used the traditional Genval workshop definition. Some studies that included patients on PPI, however, have not reported the numbers on PPI therapy at the time of endoscopy.22, 24 A study by Rosaida et al. does not report the status of PPI therapy,25 whereas a study by Ang et al. has excluded patients on PPI therapy.23 This current prospective study began enrolling its first patients in 2003 and used the Genval workshop’s definition of diagnosis of NERD. Unique to our study, we also performed a subgroup analysis on PPI-naïve patients in accordance with the Vevey definition of NERD. We found similar independent risk factors (hiatal hernia and lack of nocturnal symptoms) in both our analysis, clearly suggesting that the association is real.

It is well-known that interpretation of GERD symptoms and endoscopic findings is difficult while the patient is on PPI therapy. We had a large PPI-naïve population that allowed us to perform a subgroup analysis comparing characteristics of NERD and EO patients without the confounding effect of PPI. This has not been reported in previous studies. For the subgroup analysis, all patients who were on PPIs at the time of endoscopy were excluded from the analysis. Among these patients, there was no difference in the use of histamine 2-receptor antagonist (H2 blocker) between the NERD and EO groups (69.9% vs. 70.7%, = 0.90). Therefore, H2 blocker usage is unlikely to bias our results. We found that NERD patients were less likely to have frequent reflux symptoms (more than one episode a week), severe symptoms, nocturnal symptoms and hiatal hernia when compared with the EO patients.

A few limitations of this study should be acknowledged. There is potential for recall bias when the questionnaire was administered and the lack of asymptomatic control group limits the conclusions to some extent. The study subjects consisted mainly of Caucasian men at the Veterans Affairs Medical Center thus limiting generalizability. Due to the large number of participants in this study, it was not possible to perform pH monitoring to confirm the diagnosis of NERD in this patient population. Furthermore, acid sensitivity and microscopic oesophagitis were not evaluated in this study. GERD patients were referred to an endoscopy in case of persistent symptoms, thus introducing selection bias. No atypical symptoms were recorded on the validated GERD questionnaire. Also, a majority of the patients were on PPI therapy at the time of index endoscopy. Although this is reflective of the day-to-day clinical practice, it could potentially bias our study. Additionally, this study did not attempt to quantify the response to PPI. The current study was not designed to collect long-term follow-up data and therefore the relative incidence of BO in NERD and EO could not be determined. The strengths of this study include the large numbers of patients with GERD who were evaluated in a prospective manner and symptoms were recorded using a validated GERD questionnaire.

In conclusion, the results of this study show that PPI use prior to endoscopy can lead to significant bias in the classification of EO patients to have NERD. The odds of a NERD patient being on a PPI are three times higher when compared with EO. Lack of hiatal hernia and nocturnal symptoms were independent predictors for NERD. Further studies should evaluate biological mechanisms predisposing to NERD and EO phenotypes in patients with GERD.

Acknowledgements

Declaration of personal interests: P. Sharma has received grant support from Takeda Pharmaceuticals North America Inc., Olympus America Inc. and Barrx Medical Inc. Declaration of funding interests: This study was supported by the Veterans Affairs Medical Center, Kansas City, MO.

Ancillary