Pancreatic Acinar Metaplasia in Distal Esophageal Biopsies Is Associated With Chronic Nonsteroidal Anti-inflammatory Drug Use

The study was approved by the Institutional Review Board for the University of Texas Health Science Center at Houston McGovern Medical School.

We searched the computerized pathology database at our hospital for a 1-year period (May 1, 2016–April 30, 2017) for all patients who underwent upper gastrointestinal endoscopy with EGJ and/or distal esophagus biopsy. Four hundred forty-four consecutive patients were identified and included in this study. Copies of the pathology reports and the pathology glass slides (sections of formalin-fixed, paraffin-embedded tissue, stained with hematoxylin-eosin) were obtained, and the computerized medical records were searched for relevant information. Pancreatic acinar metaplasia was defined as previously described by Wang et al.²  The histopathologic examination was performed by a single experienced gastrointestinal pathologist. Chronic nonsteroidal anti-inflammatory drug (NSAID) use was defined as daily NSAID use for at least the 3 consecutive months leading to the date of endoscopy; chronic PPI use was defined as daily PPI use for at least the 12 consecutive months leading to the date of endoscopy.

Statistical analysis was performed using a 2-sided Fisher exact test, with P < .05 considered significant, using Prism 7 statistical software for Mac OS X (GraphPad Software, La Jolla, California). Multivariate analysis, with P < .05 considered significant, was performed using Wizard Statistical Software for the Apple Macintosh MacOS (Evan Miller; available at the app store).

Four hundred forty-four consecutive patients were identified, and the related pathology material was retrieved. Because many of the biopsies were designated “distal esophagus/GE junction” by the gastroenterologists, it was not possible to get a reliable separation by exact location. None of the lesions (PAM) were visible endoscopically. The lesion size varied on microscopic examination (M.Y., unpublished data, July 2017), but this information was not recorded/measured. The male to female ratio was 1:1.15 and the age range was 10 to 88 years (mean, 55.4 years; median, 58 years). An example of PAM is shown in the Figure. Seventy-one patients (15.9%) were found to have PAM. The presence of PAM on histopathologic examination was correlated with several clinical and pathologic findings.

There was no significant association between PAM and patient gender (24 male, 47 female versus 166 male, 207 female; P = .12), the mean number of biopsy pieces taken at endoscopy (3.5 versus 3.6, P = .75), the presence of active inflammation (9 of 70 versus 66 of 377, P = .39), Barrett metaplasia (21 of 71 versus 103 of 373, P = .58), the presence of pseudogoblet cells (9 of 71 versus 57 of 373, P = .72), or multilayered epithelium (6 of 71 versus 20 of 373, P = .28). In patients for whom the appropriate clinical information was documented, we found no significant association between PAM and clinical diagnosis of reflux esophagitis (30 of 70 versus 160 of 354, P = .79), current or previous Helicobacter pylori gastritis (12 of 71 versus 50 of 362, P = .46), history of bariatric surgery (1 of 68 versus 11 of 358, P = .70), or history of cholecystectomy (12 of 68 versus 61 of 359, P = .86).

Pancreatic acinar metaplasia was significantly associated with chronic carditis (66 of 70 versus 309 of 373, P = .01), patient age younger than 51 years (35 of 71 versus 105 of 373, P < .001), chronic PPI use (40 of 71 versus 137 of 352, P = .008), and chronic NSAID use (19 of 69 versus 34 of 347, P < .001). To determine whether the factors significantly associated with PAM were independent variables, the data were reviewed and 30 cases that did not have complete information on either or both PPI and NSAID use were removed. The remaining 414 cases, 68 of which had PAM, were used in the subsequent analysis. The results of univariate and multivariate analyses show that chronic carditis, patient age younger than 51 years, chronic PPI use, and chronic NSAID use are independently associated with PAM (Table).

In the present study, PAM was found in 15.9% of biopsies from the EGJ. This prevalence rate is slightly lower than that reported by Wang et al²  but higher than that originally reported by Doglioni and colleagues.¹  Data in the literature regarding PAM prevalence are considerably variable. Whereas Polkowski et al⁹  noted pancreatic acinar cells in 14% of cases, Sarbia et al¹⁰  found pancreatic acinar cells in 61% of specimens. It has been suggested that the prevalence rates of pancreatic acinar cells at the EGJ are highly dependent on the methodology applied for their detection, and that the prevalence rate will be lower if only one section of the biopsy is examined. In one study, it was suggested that PAM can be found in the cardiac region of all subjects if the area is exhaustively sampled.⁴ 

In a study that included 155 patients, Wang et al²  found significant association of PAM with young age and proposed that PAM is congenital in nature. In our study, although we found significant association with younger age, PAM was also significantly and independently associated with chronic inflammation at the gastric cardia and chronic use of PPIs and NSAIDs, all of which have to occur after birth. Whether a subset of PAM may be congenital needs to be confirmed by studying neonatal autopsies.

Similar to what has been reported by others,^1,11  we found that PAM is significantly associated with chronic carditis. Faller and Kirchner¹²  postulated that pancreatic metaplasia can be regarded as the result of altered morphogenesis within the gastric mucosa. Impaired expression of the gastric morphogenic factor sonic hedgehog by parietal cells and increased expression of the pancreatic morphogenic factor PDX1 seem to be crucial for the development of pancreatic transdifferentiation. Altered expression of the morphogenic factors is partly caused by changes in the gastric milieu.¹² 

Recently, experimental in vivo studies in rat models^8,13  have shown a link between long-term treatments with PPIs and risk of PAM development, and our results from this study support those reported in the rat models. In biopsies that contained oxyntic gastric mucosa, changes suggestive of PPI use, such as parietal cell hyperplasia and dilated fundic glands, were seen on microscopic examination. Proton pump inhibitors are a class of very efficient acid suppressors that are highly successful in controlling gastroesophageal reflux disease symptoms and prevent its complications, mainly esophageal inflammation and strictures. It is difficult to speculate on the mechanism by which chronic PPI use leads to PAM, especially because chronic inflammation at the cardia, which is supposed to be prevented by PPIs, is also significantly associated with PAM.

Probably the most important finding in this study is the significant and independent association between chronic NSAID use and PAM, which has not been reported to our knowledge in the English literature. Nonsteroidal anti-inflammatory drugs are a broad family of compounds primarily used to treat pain, control inflammation, and prevent heart attacks. Studies have also shown that NSAIDs are effective in the prevention of a few common cancers.^14–16  Bombardo et al¹⁷  investigated the effect of NSAIDs in the course of acute pancreatitis in terms of progression of inflammation and regeneration of the organ, using a murine model. They found that NSAIDs are able to directly inhibit mitogen-induced proliferation of pancreatic acinar cells. The findings from that study do not explain the link between chronic NSAID use and PAM in humans. To our knowledge, association between NSAIDs and PAM has not been reported in animal models.

The clinical significance of PAM at the EGJ/distal esophagus, if any, is still largely unknown. If PAM is functional and pancreatic enzymes are secreted within the local tissue of the distal esophagus or cardia, then perhaps PAM could be the cause of chronic inflammation (carditis) in that location, rather than being caused by it, which could explain why chronic PPI use has seemingly the opposite effect. It is still not known, for example, if larger PAM is associated with more significant inflammation, fibrosis, or other pathology in that location. Although no significant association is found between PAM and Barrett metaplasia in our study, we still don't know what role, if any, PAM plays in the malignant progression of Barrett metaplasia. Finally, given the divergent molecular pathways of NSAID and PPI action, it would be interesting to know if PAM associated with either drug has a different clinical implication.

In conclusion, our results confirm the significant association between PAM and chronic inflammation of the gastric cardia (chronic carditis) and the association with chronic PPI use previously proposed from a rat model. The strong association with chronic NSAID use has not been previously reported, and we believe it warrants further studies.