Three important questions were discussed: 1) What outcome measures should be used in studies of gastroesophageal reflux disease (GERD) in infants, older children, and adolescents; 2) What study design is appropriate to address questions about GERD in these populations; and 3) What new knowledge is needed?
An important question addressed by the group was whether multicenter clinical trials are needed to answer the aforementioned questions. The ethics of not doing randomized controlled treatment trials is relevant to the discussion of treating children with medications for which there are no safety or efficacy data. This approach to treatment of children could be considered “an experiment without design or informed consent.”
The definitions of esophageal and extraesophageal manifestations of GERD were discussed, but the group recognized that different specialists have varying perspectives of GERD. Outcome measures to assess GERD in children are listed in Table 1. Previously used paradigms could be applied to the GERD model. For example, the Patient Outcomes Research Team (PORT) trials addressed a number of questions regarding specific disease processes and potentially can serve as a model for evaluating disorders associated with acid reflux. In addition, a number of interagency, industry, and academic partnerships can serve as models. These partnerships exist in the areas of food safety, hepatitis C, and colorectal cancer.
For each of the following six conditions, the group recommended 1) entry criteria and 2) outcome measures, or end point variables.
ASTHMA/REACTIVE AIRWAYS DISEASE
To assess pulmonary symptoms that are related to GERD, a validated instrument scoring reproducible symptoms is needed. Investigators also can include the assessment of health care utilization, medication utilization, and functional status (i.e., quality of life, or QOL). Parents can be asked questions using a semiquantitative impairment scale (e.g., 1–5 impact on QOL). These questions can include the child's performance in school. The impact of a child's illness on the parents' QOL could be measured as both entry and outcome variables in correlation to disease severity.
Specific objective clinical criteria can be used to assess pulmonary status. Noninvasive, validated pulmonary function tests (e.g., changes in peak flow) can be performed as entry and outcome measures in the primary care or pulmonologist's office. The group's consensus was that elaborate testing is unnecessary. Recommended gastrointestinal (GI) studies would include esophagogastroduodenoscopy (EGD) and biopsy at baseline, and 24-hour pH monitoring at baseline and end point. Follow-up endoscopy would only need to be disease driven.
EROSIVE ESOPHAGITIS OR HEARTBURN
Erosive esophagitis was thought to be uncommon as a diagnosis in children, but it can occur. It was recommended that GERD-specific symptom assessment be done at entry and end point using a validated, reproducible system of scoring symptoms. The Pediatric Endoscopy Database System-Clinical Outcomes Research Initiative (PEDS-CORI) system could be used to validate a new symptom-scoring instrument at a number of centers, and then that instrument could be used in clinical trials. Clinical correlates of disease obtained by EGD or pH-metry are also needed.
Investigators could measure QOL and school performance, at study entry and end point, as functional measures of therapeutic success. Recommended GI studies would be upper endoscopy and biopsy at baseline. Follow-up endoscopic studies would be disease driven. Discussion ensued on the definition of “disease driven,” with the following conclusion: If endoscopy is performed in a child presenting with irritability or epigastric pain and erosive esophagitis is found, follow-up endoscopy in that child would be warranted. On the other hand, if EGD is performed for upper GI symptoms and no erosion is found, symptom presence or absence at study end would determine the need for repeat EGD. Thus, one could define an algorithm for when to do follow-up studies. To correlate findings with esophageal acid exposure, pH-metry is recommended at study entry and end point.
LARYNGEAL AND TRACHEAL DISEASE
The group was unanimous in its conclusion that it is critical to work with different subspecialists to develop a multidisciplinary approach to the management of extraesophageal manifestations of GERD.
With respect to symptom assessment of laryngoesophageal reflux disease, the group recommended assessment of hoarseness, stridor, and chronic cough at entry and end point. Functional assessment could be performed via evaluation of QOL and school performance at entry and end point. Direct flexible laryngoscopy and biopsy can be done fairly quickly and easily. Voice analysis also is recommended as a semiquantitative measure of disease. An easy point-of-care voice analysis test has been validated in children and can be used at entry and end point.
GI assessment would include pH-metry at baseline and end point. The otolaryngologists present in the group raised the issue of triple-probe catheters, which have been used by adult gastroenterologists in the study of extraesophageal manifestations of GERD, e.g., gastric and esophageal (lower, mid, and upper). The adult gastroenterologists have looked at different aspects of acidity, correlating findings with clinical disease. For pediatric studies, endoscopy and biopsy would be useful to correlate potential lower esophageal abnormalities with ear, nose, and throat (ENT) disorders.
DISORDERS IN NEONATES
Specific disease states that may be GERD associated include stress ulcerations, feeding intolerance, and apnea/bradycardia. Symptom assessments and pH-metry should be performed at baseline and end point.
The group felt that neurologically injured children should be considered as a distinct category. These children are difficult to evaluate and study, but a great number of antisecretory drugs are being prescribed for them.
Development of a symptom scoring or other assessment tool for this population is a difficult task. Better, more standardized entry and outcome measures are clearly needed. Use of pH-metry at beginning and end point should be considered. Endoscopy with biopsy at study entry and end point should also be considered. Assessment of dosing requirements for the resolution of endoscopic disease should be part of the outcome.
FAILURE TO THRIVE/FEEDING DIFFICULTIES
For children failing to thrive because of GER, symptom assessment, functional measures, and physical exam are appropriate, including anthropometrics (height, weight, Z scores). These patients could then be followed up after a therapeutic course of acid suppression to determine whether appropriate antisecretory therapy results in growth. These assessments could also be correlated with findings on pH-metry at entry and end point.
QUESTIONS AND ANSWERS
Comment: The problem with recommending that pH-metry and endoscopy be done before an end point is that these tests are usually not clinically indicated. Ethical guidelines and institutional review boards (IRBs) may not allow the performance in children of invasive studies that are not clinically indicated.
Comment: For most of the priority conditions, I think IRB approval to do endoscopy should be obtainable. Of course, this document is a draft in progress, and it is likely with individual study designs that those developing the study design will wrestle with these issues further.
Q: Do we as investigators need to have all this information to determine that a drug is effective? Do we need to have pH-metry and endoscopy and biopsy, or do we really want just to assess the clinical response? A good, validated questionnaire is sufficient to do the latter.
A: From an ENT standpoint, I think that pH studies and endoscopy are important because we do not know if reflux exposure has anything to do with the condition we're studying.
A: One rationale for using pH studies is that we do not know the dose of drug that will reliably and consistently de-acidify the stomach because of the variability in dosing in children.
A: I am glad our ENT colleague brought up those points. These are issues that have not been resolved in adult patients. When we met to discuss the esophageal and extraesophageal manifestations of GERD, we were thinking not only of ways that are appropriate, cost effective, and good for clinical trial design, but also of how to address some of these important unresolved scientific issues.
A: If you do not do the appropriate investigations at the beginning, then at the conclusion of the trial, you will end up with a heterogeneous group. If one wants to study patients with an atypical manifestation, for example, GER-related asthma, it is better to do pH studies to identify a group with abnormal pH. Whether you wish to do endoscopy depends on what additional information you want to obtain. In fact, you are not interested in the histology of the esophagus, because you're not treating an esophageal condition, you're treating respiratory symptoms. By measuring pH in the beginning and at the outcome, you can separate responders from nonresponders. Once an effect on acidity has been demonstrated in well-designed clinical studies, you do not need to do pH studies to treat your patients.
A: This summary started as questions. We took all the controversial issues that arose during the lectures, particularly those that different specialists have passionate feelings about, and put them in questions first. To answer some of these questions, we have to think of a careful, thorough, systematic approach that both addresses the question and is ethically sound and consistent with standard of care. It is hoped we are going to see that reflected in future study designs.
Q: The discussion of pediatric studies in atypical GERD populations is interesting because those studies address important questions and look at populations that have been ignored, but I want to ask the Food and Drug Administration (FDA) whether we need to study these patients in the context of pediatric labeling. When we look back to the adult studies in GERD, acid suppression was usually evaluated in populations with “typical” GI symptoms and manifestations.
A: I believe that clinical practice determines where drugs are needed. And if there is a perception, for example, that bedridden, neurologically difficult patients require antisecretory medication at a different dose than has been used in other settings, and it is studied in an appropriate way, then it would be appropriate to label the drug for that purpose.
Comment: From what I am hearing, some of the syndromes that are being discussed are really derivative symptoms of a primary hypothesis, which is reflux. And as they become more derivative, there are other disorders that can cause the same problem, and the differential diagnosis becomes complex. It is important when designing a clinical study not to get into an empirical treatment study in which there are multiple etiologies. You do get a dilution effect, and even if the outcome is statistically valid, it would not necessarily instruct the clinician on how to exclude the other causes, or when it may be inappropriate to use antisecretory therapy.
As a regulator, I would still be interested if the clinical outcome that was of concern was a derivative outcome, such as asthma. The outcome would still need, in some respect, to be anchored to the primary pathophysiology conceptually in the study format. That does create a problem if there are limitations in the ability to do these studies in children. It is a paradox and it is problematic.
Suggestions for possible study designs are summarized in Table 2, not in any specific order in terms of research priority. A longitudinal cohort study may address some of the issues with respect to reflux in the general population. A subcategory of the study could be disease specific, so that we can look at natural history and outcome. Another approach is the standard intention-to-treat, or per-protocol, clinical trial in study populations with a specific disease. To yield statistically significant outcomes, studies must be appropriately powered. Consensus could not be reached on the value of placebo-controlled trials in the pediatric population.
Studies need to be age-specific, and because GER in neonates may have a different pathophysiology than in older children, this group should be studied separately. Other possible groups to study are infants younger than 12 months of age, children aged 1 to 11 years, and older children aged 12 to 18 years. Group randomization or individual randomization depends on the specific condition being studied. In designing studies on esophageal and extraesophageal manifestations of GERD, exclusion of conditions that may or may not be related to gastric acid, such as eosinophilic esophagitis, Helicobacter pylori infection, otitis, and sinusitis, should be considered. Dietary considerations in the study design are appropriate because, at least in the infant age group, formula may play a significant role. Efficacy measurements should be consistent with standard of care and treatment guidelines, so when a study is being developed, standard practice should be considered. Safety data should be collected in all pediatric clinical trials. If multicenter trials are done, there should be mechanisms for ensuring standardization across sites, scientific integrity, timely reporting of data, and publishing negative results. Suggestions for new knowledge are presented in Table 3.
Q: I am wondering about an issue that might get us out of this dilemma of doing these studies over and over again in different populations, and that is the use of a surrogate. When does it become appropriate to use a surrogate, rather than a hard clinical end point? For example, when a new statin comes before the regulatory agency, the sponsor no longer has to show hard outcome data that coronary events will be reduced. If the sponsor shows that low-density-lipoprotein (LDL) cholesterol drops a significant level and there is not a safety issue, that new statin gets approved.
There are certain diseases, such as erosive esophagitis–peptic ulcer disease is another good example–for which we have absolute ironclad evidence that if you reduce the acid to a specific amount, you will heal those lesions. That is not even an issue anymore. So why, in those groups, cannot we use acid suppression as a surrogate for hard clinical outcomes? When are we going to get to the point where we can use acid suppression as an outcome, not for asthma, but for obvious diseases?
Comment: I just wanted to comment on the biopsy/pH-metry issue. We seem to be trying to balance standard of care with our scientific endeavors to answer clinically relevant questions.
The North American Society for Pediatric Gastroenterology, Hepatology and Nutrition (NASPGHAN) clinical practice guidelines for GER (section 5.2) state that “no randomized, placebo-controlled studies evaluating the efficacy of either lifestyle or pharmacological therapy for the treatment of heartburn in children or adolescents have been published. Expert opinion suggests the use of management approaches similar to those described in adult patients.” I think this opens the door for us to do those studies with available techniques. I think that statement alone, with the review of the literature that has been done, suggests that the studies do need to be done.