Latest Insights on the Relationship between Symptoms and Biologic Findings in Adults with Eosinophilic Esophagitis

Abstract

Patients with eosinophilic esophagitis (EoE) present with symptoms, such as dysphagia for solid food, accompanied by behavioral adaptions to living with this condition. These include swallowing and non-swallowing associated pain in adulthood. In children, EoE symptoms vary with age and may, among others, include vomiting, abdominal pain, and dysphagia. While the results of early swallowed topical corticosteroids trials in adults led to some controversy on the relationship between symptoms and biologic findings, studies using validated patient-reported outcome (PRO) measures have demonstrated that symptoms in EoE are associated with fairly severe biologic alterations and that patients on the low biologic severity spectrum have little to no symptoms. This non-linear nature of the relationship between symptoms and biologic findings has important implications for the length of diagnostic delay, selection of patients for clinical trials, trial duration, and long-term management of EoE patients.

INTRODUCTION

Eosinophilic esophagitis (EoE) is defined as a chronic, immune/antigen-mediated, esophageal disease, characterized clinically by symptoms related to esophageal dysfunction and histologically by eosinophil-predominant inflammation. Adults typically present with swallowing and non-swallowing associated pain and dysphagia for solid food that is accompanied by a range of behaviors associated with thes symptoms.^1,2 Children, on the other hand, experience an array of symptoms, such as vomiting, abdominal pain, and dysphagia, but these appear to vary with age and can be non-specific.^1,2

The tale of two studies and non-linear relationship between symptoms and biologic findings

Earlier literature often highlights the controversial relationship between biologic findings and symptoms in adult patients with eosinophilic esophagitis. For example, Straumann et al. have demonstrated in a randomized, placebo-controlled 15-day trial of adult/adolescent patients (36 patients, 18 patients in the experimental group, mean esophageal eosinophil count at baseline of 148 [standard deviation ± 61] per hpf) that treatment with topical budesonide improved both histologic findings and symptoms as assessed by ad hoc-developed patient-reported outcomes (PROs) instrument designed to assess dysphagia frequency and severity of dysphagia episodes (Straumann Dysphagia Instrument [SDI]).³ On the other hand, in a randomized, placebo-controlled 6-week trial of 42 adult patients with mean esophageal eosinophil count at baseline of 26 [range 12 – 89] per hpf), Alexander et al. showed that that treatment with topical fluticasone improves the esophageal eosinophilia, but not symptoms as assessed by two items of the Mayo Dysphagia Questionnaire 14-day version (validated but not specifically for EoE patient-reported outcomes instrument).⁴ In light of the current literature, it is absolutely clear that these studies epitomize two bits of puzzle that, when put together, tell us a great deal about “the controversial” relationship between symptoms and biologic findings. However, without the data from larger observational studies (more about these later), we as a community found patient selection for the studies somewhat challenging. In fact, researchers grappling with any condition, for which no disease-specific validated PRO (patient reported outcome) measures have been developed, have faced similar challenges. For example, using the data of much larger “The Study of Biologic and Immunomodulator Naive Patients in Crohn’s Disease” trial that compared three types of treatment (infliximab alone, azathioprine alone, and combined treatment with infliximab and azathioprine in 508 patients), Peirin-Biroulet et al. have shown an apparent “disconnect” between the biologic disease activity as assessed by systemic C-reactive protein, fecal calprotectin levels, and clinical activity as assessed by Crohn’s Disease Activity Index (CDAI).⁵ CDAI is a composite measure that includes, among other things, clinician-reported symptoms, laboratory findings and presence of complications. Authors attributed such disconnect to CDAI properties, namely lack of specificity of included symptoms for the diseases and lack of sensitivity of these symptoms for inflammation observed during colonoscopy. In the era pre US FDA guidance for the PROs development, IBD researchers might have included a non-specific and systemic symptoms into the clinical activity measure; however, inclusion of non-specific symptoms, albeit assessed by non-validated instruments, as end points in adult trials is less relevant to the EoE field as dysphagia, the main symptom in this patient population, is organ specific.⁶ As such, lack of disease-specific symptoms in adult trials cannot explain this disconnect between symptoms and biologic findings in adult EoE. This brings us to the second point about the sensitivity of symptoms in detecting inflammation and fibrosis, another hallmark of this disease. In ulcerative colitis, elevated calprotectin levels precede the development of symptoms indicative of flare by 3 months.⁷ At the moment, lack of non-invasive laboratory markers precludes EoE researchers from obtaining that type of data. With the Cytosponge, esophageal string test, and quantification of eosinophil degranulation urinary 3-bromotyrosine levels on the way, such data may soon shine the light on this topic in measuring more sensitive markers of active inflammation.^8,9,10 However, in the absence of these tools, there are other ways of looking at this symptom-inflammation disconnect. One such a way is by examining the relationship between symptoms and epithelial findings using data on large groups of patients with various degrees of biologic severity and symptoms (either in the context of therapy trials or cross-sectional observational studies). Until recently most trials and even observational studies were relatively small, and symptoms were often assessed either retrospectively, as simple dichotomous outcome, or using non-validated or non-disease specific instruments, which precluded the easy interpretation of the findings. In 2014, however, the Eosinophilic esophagitis community came together to develop a patient-reported outcomes measure as a part of EoE Activity Index study (EEsAI), during which endoscopic, histologic, and symptom data on 270 patients were collected prospectively.¹¹ These more recent data clearly point to the fact that adult symptoms, such as frequency of dysphagia, various behavioral adaptations to living with this condition, and swallowing associated pain, indeed correlate with biologic findings and are indicative of severe inflammatory and fibrotic alterations. For example, individuals with severe rings, intermediate- to high-grade strictures, severe exudates, and peak eosinophil counts > 320 eosinophils/mm² (roughly > 100 per high-power field) are likely to experience the kind of symptom severity that differs drastically from patients, in whom these features are either absent or present in mild/moderate form and eosinophil counts ≤ 100 per high-power field. This relationship between EEsAI PRO and biologic disease severity is mirrored by the relationship between EoE-specific quality of life as assessed by Adult Eosinophilic Esophagitis Quality of Life (EoE-QoL-A) instrument and biologic findings.¹² Along a similar vein, when Chen et al. examined the relationship between esophageal distensibility as assessed by functional luminal imaging probe (FLIP) and symptoms, they observed that that all adult patients with severe rings were likely to have a history of food impaction when compared to 20%, 25%, 39% in patients with absent, mild, or moderate rings, respectively.¹³ Authors were not able to show that severe inflammation impacts the esophageal distensibility or is associated with history of food impaction, but the number of subjects with severe exudates was fairly low in that study. In fact, if one looks at symptom severity in patients with absence of endoscopic or histologic findings, it is nearly identical to that of patients with mild endoscopic and histologic alterations. When using the data on severity of adult patients’ symptoms and biologic findings collected in the course of EEsAI study, we and other collaborators observed that symptom score of 20 points (range 0–100 points) had modest accuracy in identifying patients in endoscopic and histologic remission (little over 60%).¹⁴ In other words, EoE symptoms correlate with endoscopic and histologic activity, however, their negative predictive value is insufficient (Figure 1). Another important point that is often brought up in the context of this “controversial” relationship between symptoms and biologic findings in EoE is related to our limited ability to assess esophageal fibrotic alterations in routine studies, as FLIP technology undergoes further evolution and is not used universally. It is often argued that this vital technology may explain away some of the controversy. However, given the fact that individuals on the low end of biologic severity spectrum have very little to no symptoms, it is highly unlikely that we will be able to explain apparent lack of symptoms with this extra bit of knowledge about the biologic alterations in this disease. As such, just as in IBD, the sensitivity of EoE symptoms in detecting inflammation and fibrosis appears to be limited, and the relationship between mucosal inflammation, fibrosis and symptoms/EoE-specific QoL as assessed by PRO measures is non-linear.

Figure 1:

Eosinophilic esophagitis is characterized by a non-linear relationship between PRO and endoscopic and histologic activity. Such a non-linear relationship between PRO and biologic activity is also observed in other diseases such as Crohn’s disease, ulcerative colitis, or asthma.

The non-linear nature of the relationship between symptoms and biologic findings in adult EoE has implications for, among other things, the timing of diagnosis in relation to appearance of symptoms, patient selection, PROs use, and placebo effect in the interventional studies, and long-term care of EoE patients, which we would like to discuss briefly.

Diagnostic delay in eosinophilic esophagitis

Given the fact that low grades of esophageal inflammation can be asymptomatic yet lead to fibrosis, it is not surprising that patients with mild esophageal inflammation or fibrosis might not experience the kind of symptoms that urge the patients to seek immediate medical help. Furthermore, these findings are consistent with a long diagnostic delay, defined as time from symptom onset to the time of diagnosis, typically observed in adults with EoE. For example, adult EoE patients in Switzerland are diagnosed after a median delay of 6 years.¹⁵ The mild to moderate biologic alterations that do not readily translate into perceptible symptoms as observed in the patient is one of the factors that contribute to this long diagnostic delay in EoE patients. However, other factors, such as the fact that patients might exercise behavioral adaptations to ease living with dysphagia and mitigate the severity of this symptom, the hidden nature of the affected organ, and the relative low prevalence of this condition (and hence less than optimal awareness among physicians) contribute to long diagnostic delays in EoE.^16,17 As patients, gastroenterologists and general practitioners become more aware of EoE, it is likely that the diagnostic delay will decrease. Nevertheless, in the absence of routinely assessed and readily available biologic marker, such delay will likely never be as short as that for other inflammatory disorders, such as Crohn’s disease (diagnostic delay of 5–9 months), or those with external manifestations that patients can observe, such as joint swelling in inflammatory rheumatic diseases (diagnostic delay of 3–4 months).^18,19

Implications for future trials

The above mentioned studies about the non-linear relationship between symptoms and biologic severity have important implications for trial design in adult EoE.

For many diseases presenting with both characteristic symptoms and biologic alterations, the United States Food and Drug Administration (US FDA) clearly advocates for the use of co-primary end points with both symptoms and biologic alterations being assessed as outcomes. This type of end point ensures that patients with symptoms have objective signs of disease-specific alterations (and the other way around) and that improvement in the degree of severity of these alterations corresponds to the improvement in disease-specific symptoms. In order to be able to demonstrate a meaningful improvement in adult EoE patients, care should be taken to enroll patients with fairly severe and/or frequent symptoms (for example, EEsAI PRO score above 50) that are likely to manifest themselves in the presence of relatively obvious endoscopic and histologic alterations, both inflammatory and fibrotic in nature. However, if the medication is expected to have mostly an anti-inflammatory effect and is tested in a relatively short trial, the patients with fairly severe fibrotic alterations, such as intermediate- to high-grade strictures, should be excluded from the trial, as symptoms due to strictures in these individuals are not likely to improve over a short period of time.

In recent years, a number of studies evaluating the efficacy of different formulations of budesonide as well as monoclonal therapy with anti-IL-13 examined symptom and quality of life improvement in adult and adolescent EoE patients using validated or newly-developed EoE specific PROs symptom and quality of life measures, such as the Dysphagia Symptom Questionnaire (DSQ), Daily Symptom Diary (DSD), EEsAI, and EoE-QoL-A.^11,20,21,22 Some of these instruments have been described and extensively reviewed with respect to the presence of important measurement properties that would affect their applicability, whilst little information is available on measures that are currently undergoing validation, such as DSD.^23,24 The symptom-specific PRO measures have 24-hour recall period (7-day recall is also available for EEsAI) and assess the frequency and/or severity of dysphagia, although swallowing-associated pain and behavioral adaptations to living with dysphagia are also assessed by some of these measures. Use of these instruments combined with the better understanding of the non-linear nature of the relationship between biologic findings and symptoms in EoE, lead to recruitment of patients with more severe biologic alterations and symptoms into these trials to more likely achieve symptom response. Although two of these studies have only been published in an abstract form, only patients with mean eosinophil counts of over 70 eosinophils per hpf presumably reflecting a more severe inflammatory phenotype of the disease were recruited to all these studies. For example, in a randomized, double-blind, placebo-controlled study of monoclonal anti-IL-13 antibody (RPC4046), patients received either 180 mg of RPC4046 (n=31), 360 mg of RPC4046 (n=34), or placebo (n=34). The baseline peak esophageal eosinophil count ranged from 92 to 123 per hpf and was reduced significantly in the groups treated with the drug but not with the placebo.¹⁹ Although this study was not powered to assess symptom improvement and was fairly small, a trend towards improvement in DSD was observed in patients treated with a higher dose of this medication. Similarly, in a randomized, double-blind, placebo-controlled study by Lucendo et al., authors used budesonide orodispersible tablet (1 mg twice daily, n=59) or placebo (n=29) for 6-week treatment of adult EoE patients.²⁵ BUL 1mg BID was highly statistically superior to placebo in achieving clinico-histological remission (57.6% vs. 0%, p <0.001), histologic remission (93.2% vs. 0%, p<0.001), clinical remission (59.3% vs. 13.8%, p<0.001), and endoscopic remission (61.0% vs. 0%, p<0.001). A prolonged treatment of up to 12 weeks increased the overall cumulative clinico-histological remission rate up to 84.7%. This study should remind us not to expect a significant symptom improvement in the majority of the treated patients in the first couple of weeks but to give some time to judge symptom improvement (Figure 2).

Figure 2:

Upon introduction of an anti-eosinophil treatment, biologic activity (endoscopy and histology) tend to improve quicker than symptoms. Thus, clinical studies should evaluate PRO for a sufficient amount of time if PRO improvement is targeted as a study endpoint.

Lastly, in a double-blind, randomized, placebo-controlled study by Dellon et al., adult and adolescent EoE patients were treated with 2 mg of budesonide oral suspension twice daily for a period of 12 weeks (87 patients, 49 in the budesonide group, mean esophageal eosinophil count at baseline of 156.3 [standard deviation ± 98] per hpf).²⁶ A validated PROs symptom measure, the Dysphagia Symptom Questionnaire (DSQ), was used as co-primary endpoint. DSQ is a daily diary with the score ranging from 0 to 84 points (when examined over a 2-week period; higher scores indicate more frequent and severe dysphagia episodes). A significant difference in DSQ score decrease (of 14.3 points when compared to 7.5 points in placebo group) and esophageal eosinophilia drop were observed, when patients treated with budesonide were compared to those treated with placebo. To summarize, utilization of new generation validated instruments in these type of trials leads to a fairly consistent improvement in severity of biologic alterations and symptoms (albeit modest), even when all studies were relatively small, if patients with at least moderate gross and microscopic eosinophilic esophagitis are studied.

When one closely examines the studies in other fields, such as asthma, the improvement observed in symptoms in patients treated with drug when compared to those treated with placebo as assessed by PROs instruments including 5-item Asthma Control Questionnaire (ACQ-5, score ranges from 0 to 6 points), St. George’s Respiratory Questionnaire (SGRQ, score ranges from 0 to 100 points), or 22-item Sinonasal Outcome Test is often modest and also dependent on the degree of underlying inflammation. For example, in a study examining the efficacy of mepolizumab (anti-IL-5) in patients with severe asthma (and recurrent exacerbations, n=576), a difference in SGRQ score change from baseline to end of treatment of 6.4 and 7 points (a drop 4 points considered to be clinically relevant) was observed between two drugtreated arms and placebo was observed.²⁷ In a study of dupilumab (antibody to the alpha subunit of the interleukin-4 receptor) in patients with persistent asthma (n=104), an improvement of 0.73 points (0.5 points is the minimal clinically important difference) over placebo was observed.²⁸ As such, it appears that the kind of symptom responses that are observed in adult EoE trials are consistent to those in other internal allergic diseases, such as asthma, where there is a broad range of fibrosis and inflammation.

Although the use of comprehensive PRO symptom instruments, especially those with daily recall, appears to be a hallmark of many new therapy trials in adult EoE, it is also important to point out that the use of simpler tools, such as visual analogue scales and 10-point Likert scales has been shown to be effective in detecting response to treatment in clinical trials and observational studies or examining symptom variation in a cross-sectional study.^11,19,23,29 In fact, score derivation for EEsAI PRO was carried out using patient global assessment of symptom severity as gold standard.

Another unique trait of the budesonide oral suspension efficacy study, when looking at all adult trials in EoE, is a 4-week placebo run in period used to ensure that patients with too few episodes of dysphagia did not enter the main part of the study. However, it was also interesting to learn that 24/93 patients had ≥ 30% decrease in DSQ score (defined as symptom response) during this period.³⁰ In a randomized, double-blind, placebo-controlled study by Miehlke et al., authors used budesonide orodispersible tablet (1 mg twice daily, and 2 mg twice daily) and viscous suspension (2 mg twice daily) for 2-week treatment of adult EoE patients.³¹ Although most patients treated with various formulations of budesonide achieved histologic remission (baseline peak esophageal eosinophil count of more than 200 per mm² of hpf; more than 95% of all patients treated with budesonide achieved histologic remission of < 16 eosinophils per mm² of hpf). These patients had a symptom response as assessed by non-validated SDI that was indistinguishable from that in patients treated with placebo. Bottom line, it appears that in the first month of treatment, a fairly substantial symptom improvement is observed in patients treated with placebo effect. These data are reminiscent of the kind placebo effects observed in asthma and IBD studies, when PRO and clinician-reported instruments, respectively, are used to assess symptoms and disease-specific quality of life. For example, in the already mentioned mepolizumab trial, a profound drop in ACQ-5 (about 0.4 points) was observed not only with drug but also in placebo treated patients (mirrored by increase in FEV₁, which to some extent depends on patient compliance).²⁵ In a trial of vedolizumab (anti α₄β₇ integrin antibody), a drop in CDAI of roughly 50 points occurred in all the study groups, including placebo, for the first 30 weeks of the study (response was defined as drop of ≥ 100 points).³² As we learn more about EoE-specific PRO instruments and conduct more EoE trials (including those with placebo run-in periods), we will find out whether longer trial periods are needed to demonstrate a symptom response to therapy, as is the case in many other fields. We also need to understand better if a placebo response is purely symptomatic or biologic as well.

Long-term management of EoE

Just as patients may sustain a histologic response in the absence of symptom resolution, a lack of symptoms does not indicate lack of inflammatory and fibrotic alterations. As a result, such patients cannot be followed based on symptoms and should undergo esophagogastroduodenoscopy to monitor biologic findings. In addition, physicians seeing the patients with this disease face the ultimate challenge: encouraging asymptomatic EoE patients with sub-clinical esophageal inflammation to continue their anti-inflammatory pharmacologic therapy. However, the question arises whether physicians should encourage the patients to undergo follow-up and continue their therapy in the absence of symptoms. The answer is: yes. The retrospective analyses of the data from groups of patients from Switzerland and United States have shown that duration of diagnostic delay is positively associated with formation of fibrotic alterations, such as rings and strictures, that in the long-term cause most symptoms and are risk factors food-bolus imactions.^15,33 Furthermore, a recent study by Greuter et al. have shown that 82% of adult patients that achieved long-lasting (≥ 6 months) clinical, endoscopic, and histological remission following median 89 weeks of treatment with swallowed topical corticosteroids, experienced clinical relapse after a median 22 weeks off therapy.³⁴ Whilst this is a first study of its kind, many of the above mentioned compounds will also be evaluated in the long-term maintenance trials, which would undoubtedly shed more light on this topic. Nevertheless, EoE physicians would have to borrow a page or two from studies on untreated hypertension, diabetes, and other chronic conditions, and conduct studies on how to best motivate patients in self-reported good to excellent health to adhere to therapy given the inevitable recurrence of disease once therapy is stopped.

SUMMARY (OR TWO STUDIES REVISITED)

So what can we learn from the two studies that have triggered this major controversy of symptoms vs. histology in the field of adult EoE? Looking closely at the studies by Straumann et al. and Alexander et al., it is absolutely clear that with markedly elevated baseline eosinophilia of 135 eosinophils per high-power field (over 439 eosinophils/mm²), the Swiss study was much more likely to be defined by a patient population with severe rings, strictures, exudates, and more importantly a high enough number of dysphagia episodes per week to detect both symptomatic and histologic improvement, When compared to the Mayo Clinic study, into which patients with relatively mild inflammation were recruited. Alexander et al. showed us that patients with 26 eosinophils/hpf (83–102 eosinophils/mm²), are likely to have relatively mild to no symptoms that could be improved in a meaningful way and therefore lack evidence of symptom to histology correlation.⁴ The results of these studies provided the first evidence for the non-linear nature of the relationship between symptoms and biologic findings in adults with EoE depending on the degree of underlying mucosal inflammation. This has important implications for the diagnosis, patient selection for the trials, and long-term management of this condition. Patients with EoE are likely to be diagnosed with a delay that can be in part attributed to the lack of profound symptoms in the presence of mild biologic alterations. When selecting subjects for the trials, care should be taken to recruit individuals with at least moderately severe biologic alterations and symptoms in order to show a meaningful improvement. The use of validated PRO measures as well as simple numeric scales to assess symptoms and disease-specific quality of life in these trials is encouraged. The symptom improvement observed in most recent trials is modest (and most likely clinically meaningful) and consistent with modest symptom improvement observed in trials of other chronic conditions, such as asthma or IBD. The trials should be at least 6–8 weeks long given the symptom improvement on placebo in the first month of the trial. The lack of symptoms at low biologic severity disease spectrum introduces the need for the follow-up that included esophagogastroduodenoscopy, and presents physicians with a challenge of encouraging asymptomatic patients to adhere to therapy.

As we look into the future of symptom assessment in EoE, it is absolutely crucial that we continue to validate the PRO measures, develop definitions of response and remission, and improve their properties by, among other things, making better use of various quantitative methods to support PRO development, such as classical test theory early in the PRO development and item response theory in later stages, as this may require fairly large sample sizes.³⁵ Many of the PRO developers out of necessity, relatively low patient numbers, and time considerations, examine a single symptom of EoE - dysphagia. Although this symptom is not at all easy to assess, one may hope for the tools that to begin with incorporate quite a few symptoms of this disease, undergo item reduction, and score derivations based on these more labor intensive methods. Development of these types of measure would require the kind concerted efforts between research community and pharmaceutical industry attempted by Evidera’s Exacerbations of Chronic Pulmonary Disease Tool PRO Initiative, which resulted in the development first United States FDA-approved PRO measure.³⁶

KEY POINTS.

• Earlier studies had conflicting results on the nature of the relationship between symptoms and biologic findings.

• Based on studies using newly-validated PRO measures, the relationship between symptoms and biologic findings in adult patients with EoE is of a non-linear nature: symptoms tend to be indicative of severe biologic alterations, but lack of symptoms does not exclude presence of mild to moderate biologic alterations.

• This non-linear relationship between symptoms and biologic findings has important implications for the following:

• Length of diagnostic delay

• Selection of the patients for the trials and observational studies

• Long-term management of EoE patients