Histological Presentation of Eosinophilic Esophagitis: Simply a Magic Number?

The diagnosis of eosinophilic esophagitis (EoE) increased in parallel to the number of citations in MEDLINE (Fig. 1, 2a–d), but accurate diagnosis is also critically dependent on the number of/availability of gastroscopies with esophageal biopsies. EOE was seen as part of the spectrum of eosinophilic gastroenteritis first described by Kaijser in 1937 [1]. It took until 1994 until EOE was defined by Straumann et al. [2] and nowadays seen as a distinct entity that is confined to the esophagus with characteristic endoscopic signs and clinical symptoms. Straumann identified proper clinical history and histological workup as the key elements for identifying patients suffering from EoE and had foreseen that systematic workup would identify many more individuals with EoE in the future. As Katzka et al. [3] pointed out 20 years ago, EoE changed from an obscure finding in some specialized pediatric centers into a much more common condition than anticipated before. Straumann et al. [2] were convinced by their own analysis and later data from Attwood [4] that many individuals experiencing troublesome symptoms were not correctly diagnosed but sent to psychiatry or surgery for various operations (e.g., esophagectomy or fundoplication). The problem is that these mistreatments are still carried out in some patients. The reasons are multifactorial, partly missing knowledge mostly combined with insufficient or missing diagnostic workup that may still lead to mutilations of patients. Despite esophagectomy cases, even after operations and complete lack of symptom control, involved clinicians do not revise their attitude or clinical diagnosis that led to the wrong indication for an operation. However, it should not be overseen that some patients have overlapping symptoms of combinations of EoE with other conditions like GERD or histamine-related disorders. In some cases, the diagnosis of EoE was revised to reflux disease due to the so-called positive proton-pump inhibitor (PPI) test and patients were considered candidates for fundoplication [4] since the EoE diagnosis was overruled by the result of the PPI test. Under PPI, EoE may improve but is not cured and infiltrates are still visible. Minimum numbers of eosinophilic granulocytes range between 15 per high-power field (hpf). A more recent approach asks pathologists to give the number per mm² since the optical properties of different microscopes depend on the microscope setting. Factors between 2 and 4 have to be calculated to reach the number of cells per mm² [5]. The typical histology of an EoE is shown in Figure 3. It is well known and written in most textbooks that eosinophils do not belong to the esophageal epithelium at all [6]. In fact, not even a single cell should be present. As everywhere eosinophils are a rather unspecific reaction of the immune system in various conditions throughout the GI tract. Therefore, eosinophils in the distal esophagus in GERD are not uncommon but they do not form clusters as in EoE. On the other hand, the absence of clusters is no proof that GERD only rather than EoE is present. It is also known that eosinophils can be part of an allergic reaction to salicylic acid compounds within the GI tract.

Taking into account that a single eosinophilic granulocyte can be considered abnormal, the consensus-driven minimal numbers sufficient for the histological diagnosis seem strange, especially when there are typical symptoms like bolus obstructions or odynophagia. The endoscopic sign of esophageal rings may also be variable. It looks like the number of infiltrating eosinophils may correlate to a certain degree of severity of the eosinophilic infiltration. But there is no clear-cut correlation so far. This may also be due to the fact that infiltrations of eosinophils may present as rather patchy in a circular or longitudinal orientation. This means that a biopsy taken 0.5 cm lateral or more proximal or distal of the main focus of eosinophils may show squamous epithelium with no pathological changes at all. Furthermore, it is known that counting or estimating a certain number of cells varies among individuals in a way that the numbers may also differ a lot [7].

Thus, counting to a pseudo-exact level makes no sense but seems to fulfill a clinical desire for exact numbers. It should not be forgotten that the aim is to treat patients with symptoms and not histological numbers. The question is whether clinically a certain histologically driven number will change the therapeutic scheme with, e.g., one or more eosinophils less or more. Or should the threshold to change a therapy be 5, 7, 10, or 110 eosinophils? It is clear that once the eosinophils are gone, proven by mapping biopsies and a so-called deep remission [8] is reached, the symptoms are also vanished. However, in case one eosinophil is left, it is no true remission but just falls into the given definition of less than 15 per hpf or less than 60 per mm². Grading the severity of the disease in which numbers of eosinophils seem not to correlate with symptoms and endoscopic findings does not make much sense. Also, it has to be taken into account that other parameters like subepithelial fibrosis cannot be properly graded and that there is a huge variation in assessing the thickness of the basal cell layer where esophageal epithelial proliferation starts. The same applies to the length of the stromal papillae [7]. Both parameters do increase with proliferation. So, this constitutes another set of parameters with pseudo-exact meaning. It should also not be forgotten that the thickness of the basal cell layer and elongated stromal papillae are just indicators for enhanced proliferation which is considered a very unspecific sign. The numbers do not correlate with symptoms, and in the distal esophagus, it can be related to reflux disease [9]. So, in fact, mixing proliferative markers with numbers of eosinophils is mixing up diagnostic criteria (eosinophils) with rather unspecific markers for proliferation based on various reasons. Clinicians want pathologists to assess these parameters but it does not make much sense at all. However, there is no doubt that proposed classifications or grading systems work and are reproducible. The most cited and used grading system nowadays is probably by Collins et al. [10] with pages of criteria to be graded.

But to be honest, in clinical studies, all of these parameters and more need to be assessed to enable comparing even very subtle effects, especially in conjunction with medical therapy. Also, it is very understandable that proper inclusion and exclusion criteria are needed and that remission needs to be defined precisely. Otherwise, especially intention-to-treat studies cannot compare the further course of the included individuals. It has been proven to have some value to use a so-called central reading pathologist to minimize variations during the histological assessment. In routine diagnosis, it makes no sense to produce pseudo-exact numbers and use grading that does not have much of a clinical impact since treatment is purely symptom based but not eosinophilic infiltrate based. One further problem may be seen in the very focal accumulation of eosinophils in the form of clusters of eosinophils. Such clusters cannot be found in reflux disease, but no clusters do not mean EoE can be excluded. The morphological correlate of such clusters is the characteristic whitish dots. So, if a biopsy of such a cluster is taken, the number of eosinophils will increase to before unseen heights whereas a biopsy taken 0.5 cm lateral may not show any abnormal findings in histology at all. Therefore, the practical point is that, when there is a clinical need to produce high numbers of eosinophils, biopsies should be taken from such whitish elevations or dots if there are any present.

Even if we could use artificial intelligence for very exact numbers within a particular biopsy, it still may not make any sense because the patchiness leads to non-representative values. Furthermore, there is a tendency in guidelines to step down from at least 2 biopsies each from the upper, middle, and lower esophagus for diagnostic purposes in separate containers, to take single biopsies from the upper and lower esophagus only. This may lead to the situation that patchiness maybe even less certainly detected histologically. What is needed here with no doubt are evidence driven rules for or against biopsies but not a clinical tendency to take as few biopsies as possible so that a histological diagnosis becomes so unsure that the biopsy itself makes no sense anymore.

It all started with dietary restrictions like the basic diet by Kelly et al. [11]. Then, a modified diet avoiding, egg, fish, meat, gluten, and soja – called an elimination diet [12] – was proposed. Then the idea came up with swallowed but not inhaled steroids (in contrast to asthma where inhaled corticosteroids are used [13]. The consequence was that cases with moniliasis occurred that needed recurrent treatment against the fungal infections. The next step was the idea to turn liquid steroids into a gel-like phase [14]. Sometimes the costs in the local pharmacies led to rage at the level of health insurance that was asked for reimbursement. Some attempted the possibility of mixing the liquid steroids with ampho-moronal, an anti-fungi gel-like medication that would adhere longer to the esophageal epithelium than with swallowing alone. But still, this was not optimal since it was not clear what expiration date such a mixed medication would have and thus could not be used for mass production. The situation got better once pharmaceutical companies started to use their existing formulas of topical steroids and turned them into a gel-like phase or other application preparations that allowed prolonged contact with the esophageal epithelium.

Subsequent studies came up with several results: (a) the treatment works, the eosinophils vanish, the reactive changes disappear, (b) it is not clear yet whether the subepithelial fibrosis will also vanish, (c) the treatment seems to create an on-off situation in the majority of cases: complete remission with no sequel left, (d) it is not clear yet how long this remission will last individually, (e) there are some cases that do not achieve a remission phase. The latter cases are very interesting. It can be assumed that the patients surely took the medication as intended; therefore, resistance to steroids must be assumed. This can be primary or secondarily acquired during treatment phases. Some of these resistances can be tested and susceptibility foreseen. Unfortunately, our own unpublished studies in this field showed that only around 50–75% of the resistances can be identified. Therefore, routine testing for steroid susceptibility cannot be justified.

In summary, topical steroids in their variant galenics and preparations are combined with anti-reflux medication such as PPI, anti-histamines and often combinations of all three. It needs to be noted that especially PPI can lead to PPI exhaustion, leading to increasing dosages until these medications are not effective anymore. In such situations, we have seen patients that benefit from the group of potassium-competitive acid blockers (p-cabs) a group of reversible potassium channel blockers as an alternative for patients with PPI resistance [15].

Finally, immunotherapy has been recently added as a therapeutic option for EoE. In fact, in January 2023, the EMA approved the anti-IL-4R antibody dupilumab for EoE that suppresses IL-4 and IL-13 signaling. This new agent broadens our armamentarium to treat EoE in humans.

There is no doubt that clinical studies require as much data as possible and that intention-to-treat studies need histological data before as well as after the intended treatment. It is known that some individuals cannot cope with tablets but with other galenic preparations such as liquids or gels. This should be taken into account when prescribing one or the other form of topical steroids.

There are numerous classifications listed and validated in the literature such as the EoE Histologic Scoring System (EoE-HSS) [10], the Normalized Enrichment Score (NES) [16], the peak esophageal intraepithelial eosinophil counts (PEC) [17], or the EoE histology remission score (EoEHRS) [18]. No doubt, all of these scores work and allow the observer to get an impression of what happened histologically compared to the last set of biopsies. The problem behind all scores is that they do not provide details about the exact location of the set of biopsies other than a rough recommendation to take biopsies from the upper, lower, and middle esophagus. This may lead to a failure to explain results in some patients since the inflammatory changes may be quite patchy and focal.

The next issue is that unspecific reactive changes that are more or less a bystander to any inflammatory infiltration are graded as well. Partly, they are graded based on Ismail-Beigi et al. [19] assessing in percentages of the whole epithelial thickness (not taking into account that reactive changes may lead to a thicker epithelium) [20] or in micrometers that take the whole epithelial thickness into account. The closer these stromal papillae (containing capillaries) are to the surface, the more red an area will appear endoscopically [21]. Neither a normal value of these parameters has ever been defined in clinical studies nor the moment from when a value should be considered abnormal. In other words: it is still not clear when these values have to be regarded as abnormal or when these values should be regarded as being normalized after therapy.

There is a single study that shows what happens with these values after several weeks of PPI treatment [22]. It can be assumed that after PPI we are looking at “normalized” values, but there is no proof and no validation so far. Such facts and restrictions seem not to have found their way into scoring or grading schemes. Interestingly, eosinophilic infiltrates are always accompanied by reactive changes such as basal cell hyperplasia and elongation of stromal papillae. One cannot be observed without the other and that makes it rather questionable whether such changes should and could be scored or graded at all.

One could argue that all these scores make no sense in routine settings. But again, in clinical studies, the aim should be to collect as much data as possible and as precise as possible, always to make sure that even subtle changes can be detected with all restrictions the assessments may have.

Taking into account that values and thus scores are uncertain, that therapies including dosages are more driven by the severity of clinical symptoms, and that it is known that therapies of EoE lead to a rather on-off phenomenon after an individually varying amount of time [23], scoring/grading may be completely unnecessary in a routine setting. Clinically, it is important whether there are symptoms and any eosinophilic granulocytes, and it does not have much impact whether the eosinophils count is, e.g., 134 or 127. The sole presence of a single eosinophilic granulocyte indicates that there is a pathological situation present. So, why not report just the fact whether there are any eosinophils present or not and whether they are arranged with a uniform distribution or just patchy and focal?

Taking all of the above into account, the calculation of the eosinophils per mm² instead of hpf does not make much sense either. The background is that clinical colleagues identified (correctly) that the area covered by an objective and the ocular on a microscope varies with the optical system. This already plays a huge role when counting mitoses, e.g., in breast carcinoma and there it is directly connected to grading and prognosis. It has not been shown that this is the case in EoE.

Subepithelial fibrosis is one of the hallmarks of EoE and may lead in extreme cases to a severely immobilized esophagus with stenoses. Such a situation may require repeated dilatations. Such dilatations may lead to tears and esophageal ruptures that do not heal easily and will lead to further stenoses, a situation that can easily be described as vitious circle. Besides symptom control, such a situation should be avoided. Nevertheless, from time to time, we see such patients and a minority undergo esophageal surgery (methods ranging from repeated fundoplication or esophagectomy) with sometimes disappointing results and potentially leading to mutilations and a decrease in quality of life either due to the sequel of the operation itself or unwanted side effects that can occur in some cases.

All medical therapy aims to control or minimize subepithelial fibrosis. Unfortunately, there are some shortcomings. The first problem is that there should be enough subepithelial soft tissue attached to the epithelial biopsy to allow assessment. In the esophagus, it is extremely difficult since most biopsies shear off the apical epithelium, only with no lamina propria attached. One could go so far that with no lancet it is probably not possible to gain a sufficient part of the subepithelial tissue with a regular biopsy. The next issue is that it has never been defined what a sufficient biopsy is. Should it be as wide or as deep as possible or both? There are no data available, nor any clear-cut recommendation how to achieve it anyway [24].

The next issue is how to quantify fibrosis or at least grade it somehow. Numerous schemes are available but they all suffer from issues with validation and correlation. We still cannot identify those who will progress and those who will not. Furthermore, we do not know whether a certain degree quantified however will correlate to endoscopy and/or clinical symptoms [25].

Besides quantifying by means of thickness and density of collagenous fibers there are more sophisticated methods available such as calculation of the area covered by collagenous fibers in a chosen microscopic field by color threshold analysis. In the end, a percentage will be the result. Such results are quite reproducible on a defined optical system that means once you transfer the color thresholds for the calculation to a different laboratory or a different microscope it will produce different results, which may not be that important once established in a certain laboratory with a certain recipe for H&E routine stain, but it simply makes not much sense and no benefit for the patient can be seen at this point.

Trichrome stains can also be used for quantification and distribution of collagenous fibers. Such stainings are long established and kits are commercially available that work very well and are well reproducible in every laboratory. Again, the problem for a pathologist is how to quantify the results of such stains. It is for sure depending on the depth and width (size) of the biopsy as well. Fibrosis is a sequel of long-standing inflammation and repeated damage and healing and it is known that it can be focally enhanced leading to stenoses with potentially poor outcomes. But up till now, we are still lacking a real gold standard for all cases. In conclusion, we know what we want to avoid and we can detect it when it is there, but still, we cannot correlate it. Therefore, it is very important to collect data on the presence and absence of changes during the follow-up as good as we can in clinical studies but for sure not in routine settings. Best in this respect are well-designed intention-to-treat studies. These data can then be used for consensus conferences.

With no doubt, the number of esophageal intraepithelial eosinophilic granulocytes provides some information on the efficacy of treatment. Other histological criteria are rather dependent on morphological criteria such as a higher number of eosinophils that go in parallel with reactive epithelial changes such as increased length of stromal papillae, increased thickness of the basal cell layer, and dilations of intercellular spaces or are difficult to sample and/or grade such as subepithelial fibrosis. Please see Table 1 for details.

It also needs to be noted that esophageal epithelium with no pathological changes does not contain any eosinophilic granulocyte or other inflammatory cells, especially not in the upper and middle esophagus, that means that a cut off of 15 eosinophilic granulocytes per high-power field may be a pragmatic choice but is not reflecting reality. Furthermore, the inter- and intraobserver variation in counting eosinophilic granulocytes is very high, and thus, exact numbers may not reflect reality neither. Additionally, calculated numbers in mm² do not add more information nor offer a more precise solution to this problem. It needs to be questioned whether ranges instead of exact numbers of eosinophilic granulocytes would not be a better choice. This could be also divided into 4 categories: none, few, moderate, and numerous. But such a pragmatic and simplified approach for sure needs prospective data and a consensus that recommends such a simplified approach.

The authors would like to thank their colleagues from the Klinikum Bayreuth for stimulating discussions.

W.S.: no conflict of interest. M.V. received honoraria from FALK, Malesci, Olympus, AbbVie, Lilly, and Esocap. M.N.: advisor for Pentax, PPM, Takeda, AbbVie, Roche, Boehringer, Janssen, and Pfizer.

No funding was received for this study.

Data curation: W.S. and M.V. Conceptualization, formal analysis, project administration, resources, supervision, validation, and writing – review and editing: W.S., M.V., and M.N. Writing – original draft: M.V. All authors have read and agreed to the published version of the manuscript.