Immune-mediated Gastritis in a Patient with metastatic Lung Cancer due to Therapy with the immune Checkpoint Inhibitor Pembrolizumab – Differences and Similarities in Comparison to “endogenous” autoimmune Type A Gastritis and a review of literature

Abstract

Immune checkpoint inhibitors are increasingly used in advanced malignant diseases and are well-known for their good results. With the blockade of immune checkpoints, the probability of immune-related adverse events is also increased.

We present a 54-year-old female patient with advanced NSCLC. She was treated with pembrolizumab and developed a stable disease under therapy. After six cycles, she presented with massive epigastric pain to our emergency department. Gastroscopy showed severe erosive-fibrinous pangastritis without the involvement of the esophagus, duodenum, or other immune-related adverse effects. Histology showed the complete destruction of the gastric mucosa. We concluded an immune-mediated gastritis by pembrolizumab, after the exclusion of other differential diagnoses.

Despite treatment with prednisolone and marked improvement of her symptoms, the mucosa was never fully reconstituted into a healthy mucosa.

Furthermore, we collected published reports of similar cases and conducted a comparison with features of a typical, endogenous type A gastritis to highlight similarities and differences.

Introduction

In recent years, therapeutic options for cancer have markedly increased. The approval of immune checkpoint inhibitors (ICIs), such as pembrolizumab, marked a watershed. This monoclonal antibody blocks binding between programmed cell death-1 (PD-1) receptors on activated T-cells and programmed cell death-Ligand-1 (PD-L1) on the target tissue. This stimulates the secretion of interferon-gamma by activated T cells, resulting in higher cytotoxicity and increased secretion of pro-inflammatory cytokines ¹ . Hereby tumor cells are cleared by the immune system.

For patients with advanced non-small cell lung cancer (NSCLC), it became first-line treatment if the tumor expresses more than 50% of PD-L1 on its surface ² .

Well-known adverse events are immune-related adverse events (irAEs), particularly of barrier tissues ³ such as skin ⁴ , of the gastrointestinal tract with immune-mediated colitis ⁵ and of the endocrine system such as autoimmune hypophysitis ⁶ .

We present a case in which pembrolizumab-induced immune-mediated gastritis with the complete destruction of the gastric mucosa and definite sparing of the esophageal and duodenal mucosa. Furthermore, we discuss differences between “endogenous” autoimmune Type A gastritis and present a brief literature review of pembrolizumab-induced gastritis.

Case presentation

A 54-year-old female patient underwent treatment with carboplatin/pemetrexed/pembrolizumab for metastatic NSCLC, which is considered first-line treatment in this case ⁷ . The primary tumor was located in the right upper lobe and had metastasized to the adjacent lymph nodes, bones and, most likely, the brain. After the sixth treatment cycle, she presented to the emergency department of our tertiary care hospital with severe epigastric pain, nausea, vomiting and anorexia. She denied hematochezia, melaena or hematemesis. Clinical examination revealed umbilical and epigastric pain on palpation. No further signs of an alteration of other organ systems could be detected. Her past medical history revealed uterine fibroids and hypertension. Cardiovascular risk factors included 30 pack years. Due to bone metastases, this patient suffered from severe back pain. Regular medication included opioids and intermittent use of low-dose NSAIDs. Initial blood work did not show a significant pathology.

A gastroscopy shortly after admission and 5 days later showed diffusely hyperemic gastric tissue throughout the stomach from the cardia to the antrum ( Fig. 1 a, b, c). In addition, multiple fibrin-covered erosions were found ( Fig. 1 a, b, c). Most importantly, the esophageal ( Fig. 1 a) and duodenal mucosa ( Fig. 1 c) looked unremarkable endoscopically.

Fig. 1.

a Hyperemic gastric mucosa in the cardia and fundus with multiple fibrin-covered erosions and esophageal-gastric junction with normal esophageal mucosa. b Hyperemic gastric mucosa in the body with multiple fibrin-covered erosions. c Hyperemic antral gastric mucosa with multiple fibrin-covered erosions and gastric-duodenal junction with normal duodenal mucosa.

Histology found a completely destructed mucosa in the cardia, fundus, body and antrum with entirely absent vital epithelial glands. Furthermore, parietal cells were completely absent in the gastric body. Since the picture was so uniform, only findings in the body are shown as a representative example ( Fig. 2 ). The massive destruction of cells was restricted to the mucosa. The muscularis mucosae and submucosa were found vital and intact. Overall, these findings showed massive erosive pangastritis with completely absent mucosa. The esophagus and duodenum revealed very mild infiltration of inflammatory cells only (figures not shown).

Fig. 2.

Massive inflammation with complete destruction of the mucosa in the body with entirely absent vital epithelial glands including the complete absence of parietal cells (HE, ×100).

Helicobacter pylori , cytomegalovirus (CMV) antibodies, CMV DNA or Herpes simplex virus (HSV) DNA could not be detected within the tissue and in the serum, respectively, hence making a bacterial or viral etiology of our findings unlikely.

Gastrin levels were below detection, and, in addition, anti-parietal cell-antibodies and anti-intrinsic factor antibodies were not detected in the serum. Thus, these findings excluded an “endogenous” type A gastritis.

The patient did take low-dose NSAIDs intermittently, but a type C gastritis would present with histological hallmarks such as foveolar hyperplasia, mucin depletion in epithelial cells or smooth muscle hyperplasia ⁸ .

Treatment-wise, high-dose continuous application of intravenous proton-pump inhibitor (PPI) was given initially.

But since parietal cells in the body and fundus were histologically also found to be destructed and anti-parietal cell-antibodies and anti-intrinsic factor-antibodies were not detected, another noxious agent – pembrolizumab – had to be the causing agent. Furthermore, in concordance with the histology results, the lack of measurable gastrin levels indicated the destruction of G-cells in the antrum.

It was only then that prednisolone at a dose of 100 mg was applied twice daily intravenously for seven days followed by an oral dose of 70 mg in the morning and 30 mg in the evening in order to treat this immune-mediated erosive-fibrinous pangastritis. The administration of prednisolone was in line with current treatment options for complications caused by treatment with immune checkpoint inhibitors ^{5 6 7} . After 8 days of high-dose intravenous prednisolone, endoscopy showed incipient healing of the gastric mucosa in the gastric cardia and fundus ( Fig. 3 a) and body ( Fig. 3 b). But in the antrum, massive inflammation and complete destruction with entirely absent vital epithelial glands prevailed and the inflammation continued to be confined to the mucosa, leaving muscularis mucosae and submucosa intact ( Fig. 4 ).

Fig. 3.

a Incipient healing of the mucosa in the cardia and fundus after 8 days of high-dose iv prednisolone. b Incipient healing of the mucosa in the body after 8 days of high-dose iv prednisolone.

Fig. 4.

Continued massive inflammation and complete destruction in the antrum with entirely absent vital epithelial glands leaving muscularis mucosae and submucosa intact (HE, ×100).

After discharge, careful tapering with a prednisolone reduction of 5 mg per week until achieving a maintenance dose of 7.5 mg was recommended.

A follow-up gastroscopy was performed about 2, 4 and 6 months after discharge.

At about 2 months after discharge, the patient was adherent to prednisolone tapering and took 30 mg daily. On a numeric rating scale (NRS) up to 10, the patient reported her pain and discomfort level to be around 3 under therapy with 10 mg oxycodone three times per day. Esomeprazole was taken twice daily at a total dose of 40 mg.

Gastroscopy showed healing of the gastric mucosa, although large areas of hemorrhagic gastritis were still visible. Healing was more pronounced in the fundus and body region of the stomach ( Fig. 5 a), whereas the antrum ( Fig. 5 b) was still most affected by the inflammatory destruction.

Fig. 5.

a Continued endoscopic healing in the cardia, the fundus and body. b Wide areas of fibrin-covered erosions still seen macroscopically in the antrum.

At about four months after discharge, gastroscopy showed progress of the healing in the fundus and body of the stomach; the findings resembled mucosal gastric tissue typically seen in type A gastritis. At this time also, the antral tissue had improved endoscopically ( Fig. 6 ) and histologically ( Fig. 7 a). As expected, G-cells in the antrum were undetectable as demonstrated using immunohistochemistry with Gastrin antibodies ( Fig. 7 b). Interestingly, histology revealed that the regenerated fundic mucosal tissue showed a picture normally seen in the antrum: mucoid glands without signs of regeneration of the site-specific glandular body (“antral metaplasia”) and with parietal cells still missing ( Fig. 7 c).

Fig. 6.

Incipient healing in the antrum two months later.

Fig. 7.

a Receding inflammation in the antrum and healing commencing with vital epithelial glands and focal intestinal metaplasia (white arrows) (HE, ×100). b Immunohistochemistry with gastrin antibodies showed complete absence of G-cells in the antrum; positive controls not shown (×100). c Newly formed mucosa in the fundus showed mucoid glands without signs of regeneration of the site-specific glandular body (“antral metaplasia”) and with parietal cells still missing (HE, ×100).

Her abdominal pain reduced to 1–2 on an NRS up to 10. On the other hand, her daily opioid dose was more than doubled because of bone pain. Due to progressing bone metastases, chemotherapy had to be re-initiated. Her prednisolone dose was further tapered to 4 mg per day and the dose of esomeprazole was unchanged.

Six months after discharge, another follow-up examination took place. By then, a PET-CT showed progress in the size of the primary NSCLC, and the patient had received one cycle of paclitaxel and one application of atezolizumab, another checkpoint inhibitor. Prednisolone had also been discontinued, but esomeprazole was still taken at a dose of 40 mg daily.

Gastroscopy revealed a further improvement macroscopically (pictures not shown), but mucosal tissue still displayed fibrin-covered mucosal tissue as demonstrated histologically in Fig. 8 . Immunohistochemical studies with positive and negative controls confirmed the continued absence of antral G-cells (figures not shown).

Fig. 8.

The pyloric antrum with destructed mucosal tissue toward the left and regenerating mucosal tissue toward the right, covered with fibrin (white arrow) (HE, ×100).

Table 1 summarizes our findings of immune checkpoint inhibitor-mediated gastritis, such as by pembrolizumab, and compares them to “endogenous” autoimmune type A gastritis.

Table 1 Typical findings in “endogenous” autoimmune type A gastritis versus immune checkpoint inhibitor-mediated gastritis such as pembrolizumab.

	“ENDOGENOUS” AUTOIMMUNE TYPE A GASTRITIS	OUR CASE OF IMMUNE CHECKPOINT INHIBITOR MEDIATED GASTRITIS DUE TO PEMBROLIZUMAB
REGION AFFECTED	Entire stomach, but preferably body of stomach 9 ; sparing esophagus and duodenum	Entire stomach, sparing esophagus and duodenum
ANTI-PARIETAL CELL ANTIBODIES	Typically present 10	Absent
INTRINSIC FACTOR ANTIBODIES	Sometimes present 10	Absent
PARIETAL CELLS	Absent/reduced in numbers 9	Absent
G-CELLS	Hyperplasia 9	Absent
GASTRIN LEVELS	Increased 9	Low/below detection limits
SECONDARY VITAMIN B 12 DEFICIENCY	Present 9	Possibly in development
TREATMENT	None 9 11 12	Corticosteroids 7
PRECANCEROUS STATE	Yes	Unknown

Discussion

Our case illustrates that immune checkpoint inhibitors (ICIs) such as pembrolizumab can cause irAEs in the form of immune-mediated gastritis. This rare irAE has been described in a few case reports (compare Table 2 ). In contrast, immune-mediated, i.e., colitis and hypophysitis seem to be relatively more common and are well-known adverse events ^{1 5 6} .

Table 2 Summary of reported cases of immune-mediated gastritis (reports in languages other than English were excluded; PubMed and Medline were searched). Adapted from Hayama et al. ¹³ .

Ref	Age	Sex	Type of Tumor	Drug	Duration of Therapy	Symptoms	Endoscopic Findings	Histopathological Findings	CMV	H. pylori	HSV	Treatment
Abbreviations: Pemb.: pembrolizumab, Atez.: atezolizumab, n/a: not available, +: positive, -: negative
14	71	M	Lung adeno-Ca	Pemb.	19 cycles	Epigastric pain before meals	Stomach : multiple erosions, erythema, solid white exudate	Inflammatory cell infiltration, thickening of muscularis mucosa and granulation tissue	n/a	–	n/a	Discontinuation of Pemb.
15	61	F	Ovarian-Ca	Pemb.	4 Months	Epigastric pain, nausea, vomiting, weight loss	Stomach : complete sloughing of gastric mucosa, spontaneous bleeding	Inflammatory infiltrate, granulation tissue and complete absence of mucosal glands	–	–	n/a	IV steroids
16	61	F	Lung adeno-Ca	Pemb.	13 cycles	Epigastric pain, diarrhea	Stomach : erythematous, edematous mucosa Colon: indistinct vascular pattern, edematous changes	Stomach : apoptotic bodies, mild inflammation Colon: cryptitis, foamy histiocytes	n/a	–	n/a	Prednisolone
17	75	M	Ureter cancer	Pemb.	8 Months	Epigastric pain, nausea, anorexia	Stomach : erythematous and ulcerated mucosa	Erosion of mucosa with marked inflammatory cell infiltrate	–	–	n/a	Prednisolone
18	67	F	Melanoma	Pemb.	5 Years	Abdominal pain, anorexia	Stomach : severe gastritis Esophagus, duodenum: normal appearance	Severe atrophy of glands	n/a	n/a	n/a	Methylprednisolone, PPI
19	43	F	Melanoma	Pemb.	9 cycles	Epigastric pain, nausea, vomiting	Stomach : oozing, hemorrhagic, edematous, exfoliative mucosa Duodenum: erosions	Inflamed granulation tissue and necrotic detritus	+	n/a	n/a	First dexamethasone without improvement, followed by ganciclovir
13	68	F	Lung adeno-Ca	Pemb.	25 cycles	Anorexia, nausea, vomiting	Stomach : erythema, oedema	Erosion of mucosa, inflammatory infiltrate, micro-abscesses in fundic gland mucosa	–	–	n/a	Prednisolone
20	52	M	Melanoma	Pemb.	23 cycles	Abdominal pain, nausea, vomiting, diarrhea, melaena,	Stomach : hemorrhagic gastritis	Inflammatory cell infiltration in lamina propria and crypt epithelium	–	–	n/a	Steroids, PPI
21	60	F	Melanoma	Pemb.	9 cycles	Abdominal pain, nausea, vomiting, hematemesis	Stomach : erythematous, friable, ulcerated and hemorrhagic mucosa Esophagus, duodenum: normal appearance	Subtotal loss of glandular epithelium	–	–	–	Infliximab
22	49	F	Squamous cell carcinoma of tongue	Pemb.	7 cycles	Epigastric pain	Stomach : multiple ulcers, friable mucosa, exudate Duodenum: intact	Increased lymphocytes	n/a	–	n/a	Prednisolone, PPI
23	41	F	Melanoma	Pemb.	10 cycles	Epigastric pain	Stomach : hemorrhagic gastritis	Neutrophilic infiltration in lamina propria and gastric glands, apoptotic bodies	–	–	n/a	Prednisolone
24	75	M	Melanoma	Pemb.	12 cycles	Asymptomatic	Stomach : diffuse mucosal erythema	Lymphoplasmacytic and neutrophilic infiltration in lamina propria with cryptitis, crypt abscesses	–	–	n/a	Prednisolone (1mg/kg/day, PPI (40mg/day)
					19 cycles	Epigastric pain, nausea, vomiting	Stomach : worsened diffuse mucosal erythema	Erosion of gastric mucosa, focal thinning of glandular epithelium	–	–	n/a	Prednisolone (0.5mg/kg/day), PPI (120mg/day)
25	44	M	Melanoma	Pemb.	1 month	Dyspepsia	Stomach : hiatal hernia	Lymphocytic gastritis	n/a	–	n/a	n/a
26	66	F	Colon cancer, bladder cancer	Atez. followed by Pemb.	6 months, then 1 month	Epigastric pain, dysphagia, nausea, vomiting	Stomach : erythematous mucosa with bleeding on contact, ulcers in antrum Esophagus, duodenum: normal appearance	Mononuclear inflammatory cell infiltration in lamina propria, crypt apoptosis, apoptotic abscesses	+	–	n/a	Ganciclovir
27	44	M	NSCLC	Pemb.	n/a	Dysphagia, bloating, vomiting, anorexia	Stomach : swollen, friable mucosa with exudate Duodenum : small patches of white exudate	Stomach, duodenum : inflammatory infiltrate in lamina propria, glandular distortion and ulceration	–	–	n/a	Prednisolone

Formally speaking, a pre-existing type B or type C gastritis cannot be excluded with certainty, as the damage to the mucosa was too intense for the detection of H. pylori , and histological features within the mucosa typical for adverse effects of NSAIDs could not be seen, simply because there was no mucosa left at all. Hence, we cannot prove or disprove pre-existing type B or C gastritis but, considering all findings and the patient’s history, a relevant type B or C gastritis seem unlikely, and the histological findings are like those in other published cases that are considered irAEs of pembrolizumab.

They are described in varying degrees of similarity with our case, but a small number of cases are extremely similar in their description of macroscopic and histological findings ^{15 17 18 21} . The outstanding feature hereby is the complete destruction of the gastric epithelium with the absence of mucosal glands. Also unique is the definite sparing of histo-anatomical boundaries to the esophagus and duodenum ( Fig. 1 a and 1c).

It is remarkable that most cases of immune-mediated gastritis caused by pembrolizumab seem to affect the gastric mucosa exclusively, strictly sparing the esophagus and duodenum.

In addition, two patients treated with pembrolizumab developed CMV-associated gastritis ^{19 26} . The macroscopic appearance in those cases was described similarly to our case and other cases without CMV detection, but the treatment is significantly different, obviously.

Some cases mentioned a different anatomical distribution of the affected tissues without the gastric mucosa being the affected area only; in one case study, the transverse colon was affected by the irAE with diffuse erythema and oedema ¹⁶ . Another case is described with an affected duodenum leading to gastroduodenitis associated with pembrolizumab ²⁷ .

Other cases of pembrolizumab-induced gastritis have been described with histology distinctly different from the aforementioned cases of immune-mediated gastritis, including our own, thus leading the authors to label their cases lymphocytic rather than immune-mediated gastritis ^{24 25} .

One case mentioned histological similarities to ulcerative colitis ¹³ .

So far, few case reports described the exact same histological pattern with the complete destruction of the glandular epithelium ^{15 17 18 21} .

In one case study, gastrin levels were measured, but contrary to our case, they were normal and hence proved the presence of G-cells ²² .

Interestingly, Gaffuri et al. published a temporal process of a patient showing a potential development of the severity: this patient was asymptomatic at first, but PET scans and endoscopy showed a beginning mild inflammation. Histology showed findings resembling lymphocytic gastritis. As the patient was asymptomatic, pembrolizumab was continued. After a further seven cycles, they became symptomatic with epigastric pain, nausea and vomiting – just like our patient. By then, histology found erosions of the gastric mucosa with focal thinning of glandular epithelium, which could be a preliminary stage to the histological findings in our patient with the complete destruction of the glandular epithelium.

Table 2 summarizes published cases of immune-mediated gastritis with pembrolizumab.

As previously mentioned, classic type A, B and C gastritis could be excluded because of histology, laboratory findings and the patient’s past medical history. In contrast to “endogenous” autoimmune type A gastritis, anti-parietal cell-antibodies, as well as anti-intrinsic factor-antibodies, whose specificity for autoimmune gastritis is higher than those of anti-parietal cell antibodies ¹⁰ , were not found. Also, unlike type A gastritis, gastrin levels were not measurable. Our macroscopic findings ( Fig. 1 a, b, c) correspond to the histological findings, where we could show that not only parietal cells in the gastric body were completely destructed selectively ( Fig. 2 ), but also G-cells in the antrum were destructed ( Fig. 7 b), revealing more widespread destruction of the gastric epithelium than seen in type A gastritis. Furthermore, gastrin levels in type A gastritis are typically elevated as a result of the achlorhydria following the destruction of parietal cells. In our case, as a result of the absence of G-cells, gastrin levels were below detection limits.

Furthermore, endogenous Type A gastritis is associated with gastric carcinoma and neuroendocrine tumors (NET) ¹¹ ; hence, it is seen as a precancerous state and patients are managed differently.

To our knowledge, in other case studies (except one ²² ) reporting an irAE of gastritis, the histological samples have not been tested for the absence of parietal cells and G-cells or antibodies typically seen in type A gastritis. So, our case reports the most thorough analysis of a form of autoimmune gastritis induced by checkpoint inhibitors underlining that this type of gastritis is clearly distinct from “endogenous” type A gastritis.

However, similarities include the absence of parietal cells and the following lack of vitamin B ₁₂ , as intrinsic factor cannot be synthesized. Thus, vitamin B ₁₂ cannot be resorbed in adequate quantities in the ileum. As our patient did not have pernicious anemia and vitamin B ₁₂ levels were still within normal ranges, the destruction of parietal cells must have occurred recently. But serum levels of holo-transcobalamin were low, indicating an incipient deficiency in vitamin B _12. Bodily vitamin B ₁₂ stores may last up to years, until a deficiency manifests ²⁸ . Hence, we recommend screening those patients regularly to prevent a vitamin B ₁₂ deficiency before symptoms manifest and to substitute B ₁₂ subcutaneously.

Treatment and management of patients with irAE gastritis caused by ICIs include glucocorticoid administration to downregulate the immune system. In the published case studies, patients also received PPI. Our histological findings showed that parietal cells were absent, hence PPIs cannot exert their mode of action as the target cells are missing.

But it could be expected that under therapy with glucocorticoids, the mucosa and parietal cells begin to reconstitute and PPIs could be used in a preventative fashion. In our patient, parietal cells have not evolved even after 6 months. Mucosal tissue in the fundus and body did reconstitute but a typical site-specific glandular body with parietal cells could never be seen despite extensive and repeated sampling of the patients’ mucosa throughout the treatment. Instead, antral metaplasia developed in the healing process. Likewise, mucosal tissue in the antrum did recover as well, but gastrin-producing G-cells never re-occurred.

It remains unclear if this is unique to this case or might have occurred in other cases as well. To our knowledge, none of the published cases provided a follow-up as long as ours.

Other described treatment options for irAEs include immune modulatory agents, such as infliximab ^{22 29} . This monoclonal antibody inhibits tumor necrosis factor α (TNF-alpha). We refrained from administrating those, as the patient received atezolizumab to treat her progressing metastatic NSCLC and, most importantly, she did respond to prednisolone and her discomfort levels were minor and no longer leading.

Conclusion

This case report describes an immune-mediated gastritis, a rare irAE of checkpoint inhibitors such as pembrolizumab. At first, it may resemble endogenous type A gastritis, but there are substantial differences, such as the absence of autoantibodies against parietal cells, the destruction of G-cells and, thus, the absence of gastrin. Remarkable is the strict sparing of the esophagus and duodenum. Most importantly, management of endogenous type A gastritis and immune-mediated gastritis differs critically. Immune-mediated gastritis caused by ICIs can be treated with corticosteroids or (if feasible) other immunomodulatory agents.

Contributorsʼ Statement

Miriam Ecker drafted the manuscript. Frank Schuppert treated the patient, and performed and made images of the endoscopies available. Sandra Tebbe treated the patient throughout the time as her oncologist. Helgard Weckauf processed the histological images and made them available for publication. Frank Schuppert, Helgard Weckauf and Sandra Tebbe revised the manuscript critically for intellectual argumentation. All authors gave their final approval of the version to be submitted.