CD8 T-Cell–Predominant Lymphocytic Esophagitis is One of the Major Patterns of Lymphocytic Inflammation in Gastroesophageal Reflux Disease

We conducted an observational retrospective study. Our study group was derived from 246 patients who underwent esophageal biopsies and standard manometry testing performed as part of preoperative evaluation for Nissen fundoplication at our institution from 1998 to 2014. Patients were referred to surgical treatment due to symptoms refractory to life-style changes and proton pump inhibitors. Results of a barium esophagram (barium swallow) were available for 120 of 246 patients. We considered cases as positive for motility abnormalities if they showed abnormalities either by manometry or by barium swallow at any time before or after fundoplication. Those cases were excluded. If both manometry and barium swallow were performed, a negative result of both studies was required to establish no motility disorder. We also excluded cases with absent or insufficient squamous epithelium in biopsies and those with esophagitis other than LyE and active esophagitis consistent with reflux. The latter were patients with Candida esophagitis (n = 2), pill esophagitis (n = 1), and eosinophilic esophagitis (n = 1). Crohn esophagitis was also part of the exclusion criteria; however, no cases of Crohn esophagitis were identified. Overall, 161 patients had GERD, normal esophageal motility, and satisfactory esophageal biopsies. The study was approved by the institutional review board.

For all patients meeting inclusion criteria, we obtained demographic data as well as information pertaining to clinical diagnosis, past medical history, and endoscopic, imaging, manometry, and, where available, pH-metry findings from the files and electronic medical records.

Biopsy specimens were fixed in 10% formalin, paraffin-embedded, and stained with hematoxylin-eosin. One slide with 3 step cuts and the highest number of intraepithelial lymphocytes was evaluated per case. The following histologic features were evaluated: extent of lymphocytic inflammation (focal or diffuse); localization (peripapillary or dispersed), and number of intraepithelial lymphocytes; spongiosis (intercellular edema), basal cell hyperplasia, and elongation of stromal papillae; and the presence of intraepithelial eosinophils and neutrophils. Lymphocytes were deemed peripapillary when they centered on a stromal papilla and there was an obvious transition of the density of lymphocytes from the peripapillary area to the interpapillary area. Lymphocytes were considered dispersed when they were randomly scattered in the epithelium with no relation to stromal papillae. The criteria for LyE included increased numbers of peripapillary lymphocytes and rare or absent intraepithelial granulocytes in at least 1 biopsy fragment. As squamous hyperplasia is not a common feature in LyE, another criterion of LyE in this study was the absence of squamous hyperplasia, the latter defined as simultaneous presence of basal cell hyperplasia and elongation of stromal papillae (see below) in the area of increased lymphocytes.^1–3  A single peripapillary lymphoid infiltrate was sufficient for the diagnosis of LyE. The cutoffs for a normal number of intraepithelial lymphocytes evaluated in hematoxylin-eosin–stained slides at different levels, such as gastroesophageal junction, distal esophagus, and midesophagus were 62, 46, and 41 lymphocytes per high-power field, respectively. These numbers represent means +2 SDs of the counts previously reported in healthy volunteers⁸  (the cutoff >20 lymphocytes per high-power field used for abnormal in some previous studies is not based on evaluation of healthy individuals). Rare intraepithelial granulocytes were arbitrarily defined as no more than, on average, 1 granulocyte per 2 fields of view (×400). Cells were counted in one mostly affected high-power field using an Olympus BX 41 microscope with a field number 22 eyepiece, resulting in a 0.237 mm² field of view (×400). Increased lymphocytes were considered focal when present in less than 50% of the biopsy fragment and diffuse when present in 50% or more. Spongiosis (intercellular edema) in the area of increased lymphocytes, a common feature of epithelial injury in LyE,^2–4  was defined as either irregular rounded dilatation or even widening of the spaces between neighboring epithelial cells. It was considered prominent when the space between the epithelial cells was equal to or greater than the width of the epithelial cell nucleus and it was considered moderate when the intercellular space was less than the width of the epithelial cell nucleus. Diagnostic criteria of RE consisted of the presence of squamous hyperplasia and intraepithelial granulocytes in the context of clinical GERD. Squamous hyperplasia constituted the minimal criterion of RE and was defined as the simultaneous presence of basal cell hyperplasia (expansion of the basal zone more than 15% of the thickness of the squamous epithelium) and elongated stromal papillae (lengthening to more than two thirds of the thickness of squamous epithelium).¹⁵  Two authors (RM and ML) analyzed the biopsies independently. Discrepancies were resolved by consensus reached during joint microscopic evaluation.

Routine CD4 and CD8 immunohistochemistry was performed using Bond Polymer Refine Detection staining reagents (Leica Biosystems Newcastle Ltd, Newcastle upon Tyne, United Kingdom) and Bond III autostainer (Leica Microsystems, Buffalo Grove, Illinois). The following primary antibodies were used: anti-CD4 antibody at a 1:100 dilution (clone SP35; Cell Marque, Rocklin, California) and anti-CD8 Bond TM ready-to-use antibody (clone 4b11; Leica Biosystems). CD4-positive T cells and CD8-positive T cells were counted in the same field of view (×400) where lymphocytes had been counted. The CD4:CD8 ratios of greater than 1 or 1 or less indicated predominance of CD4 or CD8 T cells, respectively. To confirm the accuracy of the results showing almost universal CD8 T-cell predominance of LyE in GERD cases (see below), CD4 immunohistochemistry was performed twice with appropriate controls.

The distributions of baseline characteristics were compared between the groups using a 2-sample t-test for normally distributed continuous variables and the Fisher exact test for categoric variables. Statistical analyses were performed using GraphPad Prism (GraphPad Software Inc., La Jolla, California). The P value was considered significant when it was equal to or less than .05. Data are presented as means ± SDs.

The study group consisted of 161 patients with GERD and normal esophageal motility (age 49 ± 12 years; male to female ratio = 80:81). Thirteen percent (21 of 161) of these patients had a history of Barrett esophagus and 25% (41 of 161) showed evidence of endoscopic esophagitis. Histologic features of RE were present in 48% (77 of 161). The vast majority of the patients received proton pump inhibitors.

Lymphocytic inflammation was present in 13.7% (22 of 161) of patients with GERD and several patterns were observed. LyE was present in 6.8% (11 of 161) of patients (age 50 ± 12 years, male to female ratio = 6:5) (Table 1; Figure 1, A through C); it was focal in 91% (10 of 11) of patients and diffuse in 9% (1 of 11). Seven of 10 patients with focal LyE showed a single peripapillary lymphoid infiltrate per total biopsy specimen. Prominent spongiosis in the area of LyE was present in 55% (6 of 11) of patients. Focal mild basal cell hyperplasia in the area of LyE was present in 36% (4 of 11) of patients and elongated stromal papillae were present in 18% (2 of 11) (Table 1). None of these cases, however, showed changes consistent with squamous hyperplasia defined as simultaneous presence of basal cell hyperplasia and elongation of stromal papillae. In a different area of a biopsy fragment or in another biopsy fragment of the specimen, features of RE were observed in 45% (5 of 11) of the cases with LyE. In 18% of patients (2 of 11), RE was present in the same tissue fragment, and in 27% of patients (3 of 11), it was present in a different tissue fragment.

The second major pattern of lymphocytic inflammation consisted of increased numbers of dispersed lymphocytes. This pattern was observed in 5.6% (9 of 161) of patients with GERD (age 49 ± 12 years, male to female ratio = 7:2) (Figure 2, A through C); it was diffuse in 67% (6 of 9) of cases and focal in 33% (3 of 9) (Table 1). In all 9 cases, dispersed lymphocytes were associated with features of RE, such as basal cell hyperplasia in 100% (9 of 9) of patients; elongated stromal papillae in 67% (6 of 9); and intraepithelial granulocytes in 78% (7 of 9) (Table 1). In 22% (2 of 9), dispersed lymphocytes were the only type of inflammatory cells present, and prominent spongiosis was not observed in any of these cases. Elsewhere in the biopsy specimen, additional areas of RE without associated lymphocytosis were present in 67% (6 of 9) of cases, a proportion statistically similar to that of patients with LyE (P = .66).

An additional minor pattern of lymphocytic inflammation consisted of increased peripapillary lymphocytes in the area of RE characterized by squamous hyperplasia (Figure 2, D through F). It was observed in 1.2% (2 of 161) of patients with GERD.

Patients with LyE and dispersed lymphocyte patterns of inflammation did not differ significantly in age, sex, site of biopsy, prevalence of GERD-related endoscopic features (eg, Schatzki ring, irregular Z-line, and Barrett esophagus), prevalence of abnormal pH-metry, or prevalence of intestinal metaplasia (Table 1). However, patients with LyE pattern had a significantly lower prevalence of hiatal hernia, endoscopic esophagitis, and histologic RE (36%, 8%, and 45%, respectively) compared with those with dispersed lymphocytes (89%, 56%, and 100%, respectively; Table 1). These findings raise the possibility that patients with LyE demonstrate less severe reflux-mediated mucosal injury than patients with dispersed lymphocytes pattern.

We next evaluated the immunophenotype of lymphocytes using anti-CD4 and anti-CD8 immunohistochemistry. CD8 T cells significantly outnumbered CD4 T cells in 91% (10 of 11) of patients with LyE, all patients (9 of 9) with dispersed lymphocytes pattern, and both patients with increased peripapillary lymphocytes in the area of RE pattern. Only 1 of 11 patients with LyE (9%) demonstrated CD4 T-cell predominance (Table 2; Figure 1, B and C; Figure 2, B, C, E, and F). Even after excluding 5 cases where LyE overlapped with RE, CD8 T-cell predominance was present in 6 of 6 (100%) of the remaining cases of LyE. These findings suggest that lymphocytic inflammation in GERD, including LyE, is characterized by CD8 T-cell predominant lymphocytic infiltrate.

Multiple 2 to 7 biopsy sets spanning the period of 5.3 ± 2.2 years (range, 2–7 years) were present in 54% (6 of 11) of GERD patients with LyE. Four of 6 patients had intestinal metaplasia/Barrett esophagus, which may explain multiple endoscopies in these patients. Two patients had a follow-up biopsy that also demonstrated LyE. In patients with a pattern of dispersed lymphocytes, 2 to 3 biopsy sets spanning the period of 4.3 ± 3.9 years (range, 2–6 years) were present in 56% (5 of 9). Two of 5 patients had a follow-up biopsy that demonstrated the same dispersed lymphocytic pattern of inflammation.

In an effort to clarify the relationship between GERD and LyE and to place LyE in the broader context of lymphocytic inflammation in GERD, we analyzed the spectrum of patterns of lymphocytic inflammation in patients with GERD. These patients were free from esophageal motility disorders/abnormalities, as the latter may be associated with LyE potentially confounding the interpretation of the relationship between GERD and LyE.^4,8  We found nearly 14% of patients with GERD had lymphocytic inflammation in the esophageal epithelium. This inflammation presented as several histologic patterns. The most frequently seen patterns were LyE (observed in 6.8% of patients and typically focal) and dispersed lymphocytes associated with RE (observed in 5.6% of patients and typically diffuse). A third minor histologic pattern presented as increased peripapillary lymphocytes in the area of RE was observed in 1.2% of patients. The feature distinguishing this pattern from LyE is the presence of peripapillary lymphocytes in the area of RE characterized by at least squamous hyperplasia, which is not a feature of LyE (see Histologic Evaluation in Materials and Methods).

In 45% of the patients with LyE, areas of RE were seen elsewhere in the biopsy specimen (ie, separately from the areas showing LyE), thereby, indicating that LyE frequently coexisted with RE. A similar co-occurrence of LyE with RE and their correlated increase in prevalence after ablation has also been reported in patients with Barrett esophagus.⁷  These findings support the view that LyE is another type of GERD-associated inflammatory process in addition to RE.

CD8 T cells outnumbered CD4 T cells in all but 1 case of lymphocytic inflammation in patients with GERD, consistent with previously reported elevated levels of intraepithelial CD8 T lymphocytes in GERD-related esophageal inflammation.^16–19  Our results build upon previous observations by establishing the importance of CD8 T-cell predominance for LyE associated with GERD. To our knowledge, no other clinical conditions have been reported in association with CD8 T-cell–predominant LyE. Although lichenoid esophagitis and treated eosinophilic esophagitis may demonstrate CD8-predominant lymphocytic infiltration, the latter does not appear to be consistent with LyE pattern. In contrast, a number of studies have reported an association of CD4 T-cell–predominant LyE with primary motility disorders, Candida esophagitis, and Crohn disease, attesting to the variability of the CD4/CD8 immunophenotype of LyE.^4,8,20,21  It is plausible that differences in CD4 versus CD8 predominance reflect distinct types of immune response to mucosal injury; however, the mechanisms of these differential responses remain to be elucidated. A possible early insight is recent data showing that Th1 cells, a subset of CD4 T cells that may be involved in organ-specific autoimmunity, are expanded in achalasia-associated CD4-predominant LyE.²²  Taking into account that LyE is morphologically nonspecific, our data suggest an association with GERD when LyE shows CD8 T-cell predominance.

The available follow-up data demonstrates that GERD-associated LyE may be chronic, as 2 of 6 cases with multiple biopsy sets demonstrated LyE at least 1 year apart from the index biopsy. That only a third of cases with follow-up biopsies demonstrated LyE repeatedly can be explained, at least in part, by mainly focal nature of LyE, the lack of specific endoscopic features of LyE, and random biopsy sampling.

The clinical significance of LyE in GERD is unclear. In our study, patients with LyE had a significantly lower prevalence of hiatal hernia, endoscopic esophagitis, and histologic RE elsewhere in the biopsy specimen compared with patients with dispersed lymphocytes pattern. These findings raise a possibility that patients with GERD-associated LyE experience lesser reflux-mediated mucosal injury. However, full assessment of the severity of GERD requires evaluation of the severity of symptoms. This information was not available to us, and further studies are needed to assess the significance of LyE in GERD.

Taking into account the multiplicity of histologic patterns, several phases of lymphocytic response to reflux injury may be hypothesized, with LyE possibly occurring at an earlier phase. As yet, however, there is little evidence to support this model. Dunbar et al²³  evaluated early response to reflux injury in 12 patients with GERD successfully treated with proton pump inhibitors. Abrupt discontinuation of therapy led to intraepithelial lymphocytosis without granulocytic response. Although patterns of lymphocytic inflammation were not formally evaluated, a supplemental immunohistochemistry figure showed aggregation of T lymphocytes around stromal papillae compatible with a LyE pattern.²³ 

The limitations of our study are that it is retrospective and that the biopsy sampling was not optimized for our specific goals, potentially introducing bias. In addition, because our study group included only patients with GERD significant enough to lead to a referral for Nissen fundoplication, it is unclear if our findings are generalizable to patients with milder degrees of reflux. Finally, owing to the relatively small number of evaluated cases with lymphocytic inflammation, larger studies are warranted to validate our findings.

Nonetheless, we believe that our findings are clinically relevant suggesting that LyE is one of the major patterns of lymphocytic inflammation associated with GERD. CD8 T-cell–predominant immunophenotype may be a useful marker of GERD in the differential diagnosis of LyE.

The data presented in this manuscript were in part generated through the Department of Pathology Translational Research Shared Resource Laboratory of the Geisel School of Medicine at Dartmouth, the Dartmouth Hitchcock Medical Center and the Norris Cotton Cancer Center.