Mucocutaneous Manifestations in Autoimmune Gastritis: A Prospective Case-Control Study

Abstract

Objective:

Autoimmune gastritis (AIG) is associated with nutritional deficiencies, autoimmune diseases and gastric malignancies. The aims of the study were to test the hypothesis that mucocutaneous (MC) manifestations occur more often in patients with vs. without AIG and to delineate patterns of MC manifestations in AIG.

Methods:

A single-center, prospective 2:1 case-control study was conducted. Cases were patients with the diagnosis of AIG based on consistent serologic and histologic findings. Controls had a normal gastric biopsy. MC manifestations were independently evaluated by 3 experienced dermatologists. We conducted a multivariable logistic regression model adjusted for age, sex, Helicobacter pylori, tobacco use and alcohol consumption in order to estimate the association between AIG (vs. no AIG) and MC manifestations, which was presented as adjusted odds ratio with 95% confidence interval (aOR-95%CI).

Results:

We prospectively enrolled 60 cases and 30 controls (mean age 53.5±15.8 vs. 53.4±14.5 years; 75% vs. 73.3% female). The pooled prevalence of MC immune-mediated diseases was higher in patients with vs. without AIG [66.7% vs. 23.3%; aOR 12.01 (95%CI: 3.51–41.13)]. In patients with AIG, seropositive vs. seronegative intrinsic factor antibodies more often had concomitant immunological diseases with MC manifestations (100% vs. 58.5%; p=0.016). The most common MC immune-mediated diseases in AIG were Sjögren’s syndrome (n=5, 8.3%), alopecia areata (n=5, 8.3%) and vitiligo (n=4, 6.7%). Nutritional deficiency-related MC findings, mainly xerosis, lingual and nail disorders, were also more common in AIG.

Conclusions:

This is the first comparative study specifically designed to evaluate MC manifestations in AIG. We demonstrated that AIG is more frequently associated with both immune- and nutritional deficiency-related MC manifestations, which might have both diagnostic and therapeutic clinical implications.

INTRODUCTION

Autoimmune gastritis (AIG) is a chronic immune-mediated inflammatory disease that affects the gastric body and fundic mucosa. In AIG, an aberrant autoimmune response mediated by anti-parietal cell antibodies (PCAs) and anti-intrinsic factor antibodies (IFAs) targets gastric parietal cells.[1] Chronic inflammation leads to the development of mucosal atrophy, metaplastic glands and predisposition to gastric malignancies.[2, 3] Parietal cell loss may also lead to nutritional deficiencies due to achlorhydria/hypochlorhydria and lack of intrinsic factor. The diagnosis of AIG relies on histologic confirmation with gastric biopsies obtained from esophagogastroduodenoscopy (EGD), with atrophy of the gastric body often with sparing of the antrum a defining feature; however, clinical correlation with serology and laboratory tests is recommended to secure the diagnosis.[4, 5]

There are limitations to our understanding of the true prevalence of AIG since it is not a condition that is screened for unless lab derangements or symptoms are identified. Notwithstanding, the estimated prevalence of AIG is 0.5–4.5% globally,[6] with some studies, including those from Chile, suggesting an even higher prevalence.[7] Regardless of the population, AIG is 3 times more frequent in woman and is more common in ages greater than 50 years.[8]

AIG represents a disease spectrum and because of the nonspecific clinical presentation, substantial diagnostic delay is not uncommon,[9] Clinical manifestations of AIG are varied and include asymptomatic disease, nonspecific upper digestive symptoms, hematological deficiency manifestations, neuropsychiatric symptoms, and increased frequency of other autoimmune diseases, especially thyroid disease. Pernicious anemia (PA), characterized by vitamin B12 deficiency, is generally considered to be the most advanced clinical presentation and occurs late in the disease.[10] The most feared complication, however, is the increased risk of gastric tumors, including gastric neuroendocrine tumors/carcinoids and likely gastric adenocarcinoma.[11] Some mucocutaneous (MC) manifestations in AIG might have immune-mediated underpinnings. Indeed, while dermatological manifestations are well-established and documented in other immune-mediated gastrointestinal diseases such as ulcerative colitis, Crohn’s disease and coeliac disease,[12] there are currently no studies in the literature that specifically evaluate MC manifestations among patients with AIG.

Given the many extra-gastric manifestations in AIG, we hypothesized that AIG may have associated manifestations in skin, mucosa or cutaneous annexes that are immune-mediated or perhaps associated with nutritional deficiencies. The aim of this study was to clinically characterize and describe the MC manifestations of patients with AIG, compared with controls without AIG.

METHODS

Study design and population

We conducted a prospective single-center case-control study in a 2:1 ratio (sample size calculation detailed below). Identification of cases: Subjects who were at least 18 years old, who had a confirmed diagnosis of AIG and were followed at the Department of Gastroenterology or Digestive Endoscopy Center at the Hospital Clínico Universidad Católica de Chile were identified. Between August 2019 to February 2020, eligible participants were contacted by email or telephone and invited to voluntarily participate in the study. The diagnosis of AIG was based on consistent serological (positive PCAs or IFAs), endoscopic and histological criteria (see also below). Diagnosis of AIG is not based solely on the presence of positive PCAs or IFAs, given their imperfect sensitivity/specificity for the diagnosis.[13] As such, we decided a priori to include patients with endoscopic and histologic findings consistent with AIG but with negative PCAs or IFAs in the primary analysis, but to perform a secondary analysis analyzing patients with AIG with vs. without consistent serology; as expected, patients with seronegative AIG comprised a small proportion of the cohort (n=5). Identification of controls: Patients who underwent EGD with biopsies according to the Updated Sydney Protocol (as described below) and without diagnostic serological nor histological evidence of AIG, including specifically the absence of atrophic gastritis (Operative Link for Gastritis Assessment, OLGA stage 0), comprised the control group (n=30). Patients with atrophic gastritis due to Helicobacter pylori (H. pylori) infection, prior history of gastric adenocarcinoma and complete or partial gastrectomy were excluded, as were women who were pregnant or lactating.

All patients provided written informed consent. This study was performed in accordance with the ethical standards enunciated in Helsinki Declaration and it was approved by the Ethic Committee of Hospital Clínico Universidad Católica de Chile (ID 190614035).

Esophagogastroduodenoscopy and gastric biopsies

All patients underwent EGD with collection of gastric biopsies during the study period at our endoscopic center. Endoscopic units were equipped with either Olympus (GIF-H190/GIF-H170) or Fujinon (EC-600ZW) endoscopes. Gastric mapping biopsies were collected following the Updated Sydney System biopsy protocol.[14] With a standard size forceps five gastric biopsies were obtained: two from the antrum (lesser and greater curvature), one from the incisura and two from the body (lesser and greater curvature). Tissue samples were stored in a custom design cassette to ensure anatomical location identification. Additional endoscopic findings were documented.

Histological analysis

Histological evaluation was performed independently by two expert pathologists. Each pathologist evaluated ~50% of gastric biopsies. Samples with uncertain findings were reviewed by both pathologists for consensus. All gastric biopsies were assessed for chronic atrophic gastritis with or without gastric intestinal metaplasia or more advanced findings. AIG histological features were defined as the presence of inflammatory infiltrate associated with mucosal atrophy (0=absent, 1=mild, 2=moderate, 3=intense), with or without intestinal metaplasia, involving the gastric body, but with absence of mucosal atrophy and otherwise normal findings or only mild inflammatory infiltrate in the antrum. Hyperplasia of enterochromaffin-like cells was an additional criterion to support histological AIG diagnosis. Biopsies of control patients were evaluated with OLGA system to exclude any degree of chronic atrophic gastritis.[15, 16]

H. pylori infection was assessed by Giemsa staining in all gastric biopsies. The presence of H. pylori did not rule out AIG diagnosis. Nevertheless, atrophy involving predominantly the antrum was attributed to this infection and these patients were excluded from the AIG group.

Serum anti-parietal cell antibodies, anti-intrinsic factor antibodies and laboratory profile

Serum samples were obtained from patients with AIG at the time of study inclusion. Serum PCAs and IFAs were measured using ELISA kit (Aesku Diagnostics, Wendelsheim, Germany; positive value >18 U/mL) at Clinical Laboratory at Hospital Clínico Universidad Católica following manufacturers recommendations. In addition, we reviewed the laboratory records at our center for laboratory tests performed in study participants, including both cases and controls, between January 2018 to February 2020. The most updated autoimmune serologies, and the lowest historical value of hematocrit, serum vitamin B12 and vitamin D, ferritin and transferrin saturation in the analyzed period, were recorded for both groups.

Patients characteristics and mucocutaneous evaluation

Demographic background and comorbidities were established by clinical interview and examination in addition to review of the institutional clinical record. Three experienced dermatologists performed a general and specific medical history interview of all patients, and as well administered questionnaires assessing prior mucocutaneous, gastrointestinal and immune-mediated diseases, history of anemia, nutritional deficiencies, nutritional supplementation, neuropsychiatric symptoms and sicca symptoms. Regarding diseases with MC manifestations, as long as patients met diagnostic criteria according to standardized dermatology evaluation at any time point, this was recorded as a positive history—that is, patients did not necessarily need to have active MC lesions; we selected this definition a priori since the presence of active MC lesions may vary according to the stage in the natural history and external factors such as medical treatment or diet.

These three dermatologists performed a standardized, non-blinded and topographically structured physical examination focused on skin, mucosa and cutaneous annexes on all patients in the study. In cases of diagnostic uncertainty of a cutaneous or mucosal lesion, a fourth dermatologist was consulted to adjudicate the diagnosis. The most relevant MC findings at the physical examination were photographed with consented authorization. Dermoscopy was performed with Dermlite DL4 dermatoscope (3Gen, San Juan Capistrano, CA) as appropriate. MC findings at the physical examination were grouped in the following eight categories: I. Pigmentary, II. Alopecia, III. Sicca, IV. Oral cavity, V. Nails, VI. Other mucocutaneous immune-mediated diseases, VII. Infectious diseases and VIII. General cutaneous findings. Fungal infections were excluded from the “nails category”, because they had no direct pathophysiological explanation, as well as acrylic or enameled nails that made examination impossible.

Statistical analysis

MC lesions, comorbidities and laboratory findings were compared between cases and controls. Three pooled variables according to the investigational hypotheses and objectives were constructed: I. Pooled immune-mediated diseases; II. Pooled mucocutaneous immune-mediated diseases and III. Investigational hypotheses (variables detailed in Table 3). Since a high prevalence of Hashimoto’s thyroiditis has been reported in the Chilean population, and the association between thyroid diseases and AIG is already well-established, grouped variables I and II were calculated including and excluding immune-thyroid diseases to evaluate this possible confounding effect.[17, 18]

Table 3.

Frequency of presentation of patients with pooled immune-mediated diseases.

Variables	Autoimmune gastritis (n=60)	Controls (n=30)	Odds ratioa	95%CI	p value
I. Pooled immune-mediated diseases n (%)b
Including immune thyroid diseasec	42 (70.0)	8 (26.7)	11.67	3.49–39.05	<0.001
Excluding immune thyroid disease	23 (38.3)	4 (13.3)	5.55	1.53–20.17	0.009
II. Pooled mucocutaneous immune-mediated diseases n (%)d
Including immune thyroid disease	40 (66.7)	7 (23.3)	12.02	3.51–41.13	<0.001
Excluding immune thyroid disease	18 (30.0)	3 (10.0)	4.75	1.18–19.16	0.029
III. Investigational hypotheses n (%)e	15 (25.0)	2 (6.7)	6.82	1.27–36.72	0.025

^aOdds ratio adjusted for sex, age, H. pylori, tobacco use and alcohol consumption. Significance p<0.05.

^b“Pooled immune-mediated diseases”: vitiligo, alopecia areata, lichen planus, Sjögren’s syndrome, urticaria, lupus, dermatomyositis, scleroderma, morphea, lichen sclerosus, vasculitis, panniculitis, psoriasis, hidradenitis suppurativa, immune arthritis, type 1 diabetes mellitus, adrenal insufficiency, coeliac disease and polymyalgia rheumatica.

^cImmune thyroid disease: Hashimoto’s thyroiditis and hyperthyroidism.

^d“Pooled mucocutaneous immune-mediated diseases”: vitiligo, alopecia areata, lichen planus, Sjögren’s syndrome, urticaria, lupus, dermatomyositis, scleroderma, morphea, lichen sclerosus, vasculitis, panniculitis, psoriasis and hydradenitis suppurativa.

^e“Investigation hypothesis”: vitiligo, alopecia areata, Sjögren’s syndrome and coeliac disease.

Sample size was calculated considering a type I error of 5% and power of 80%. For this purpose, expected frequencies of “Investigational hypotheses” (vitiligo, coeliac disease, alopecia areata or Sjögren’s syndrome) was compared between study groups. According to the literature this definition was expected to be found in ≥ 25% vs. 1% in AIG patients compared to the control group.[19–22] Thus, under these parameters, a minimum of 40 subjects in the case group and 20 in the control group would be needed. The target sample size was increased to 90 subjects (60 cases and 30 controls) to increase the ability to detect differences.

Categorical variables were described as percentages and continuous variables as means with 95% confidence intervals (95%CI) or medians with interquartile ranges (IQR), as appropriate. Age was also presented as mean and standard deviation. Chi-square test was used to compare categorical variables between groups. Student t-test and Mann-Whitney U test were used to compare continuous variables between groups depending on parametric vs. nonparametric distribution, respectively. Logistic regression models were used to estimate the association between MC alterations, comorbidities and laboratory findings between groups and were presented as adjusted odds ratios (aOR) and 95%CI. Models were adjusted for sex, age, H. pylori (based on gastric biopsy), tobacco use and alcohol consumption. Statistical significance was defined a priori as a p-value <0.05. Statistical analyses were conducted using STATA v14.2 (Statacorp, College Station, TX, USA).

RESULTS

Patients characteristics

A total of 90 patients were included in the study, 60 cases with AIG and 30 controls without AIG. The mean age overall was 53.5 ± 15.3 (95%CI: 50.3–56.7) years and 74.4% were female. There were no statistically significant differences in age (p=0.981) and sex (p=0.864) between cases and controls. Complete baseline characteristics according to study group are summarized in Table 1.

Table 1.

Demographic features and medical history by study groups.

	Autoimmune gastritis cases (n=60)	Controls (n=30)	p valuea
Age mean ± DS (IQR)	53.5 ± 15.8 (49.4–57.6)	53.4 ± 14.5 (48.0–58.8)	0.981
Sex n (%)
Female	45 (75.0)	22 (73.3)	0.864
Male	15 (25.0)	8 (26.7)
Esophagogastroduodenoscopy indication n (%)
Cancer screening	7 (11.7)	15 (50.0)
Abdominal pain	17 (28.3)	6 (20.0)
Gastroesophageal reflux	6 (10.0)	7 (23.3)
Anemia	27 (45.0)	0 (0.0)	<0.001
Other	3 (5.0)	2 (6.7)
Drugs n (%) b
Proton pump inhibitor	26 (43.3)	20 (66.7)	0.037
Metformin	14 (23.3)	6 (20.0)	0.720
Vitamin B12 supplement	48 (80.0)	4 (13.3)	<0.001
Iron supplement	16 (26.7)	3 (10.0)	0.068
Vitamin D supplement	25 (41.7)	9 (30.0)	0.282
Alcohol	32 (53.3)	15 (50.0)	0.765
Smoking	13 (21.7)	13 (43.3)	0.033
First-degree family history n (%)
Any immune-mediated disease	19 (31.7)	9 (30.0)	0.872
Mucocutaneous immune-mediated disease	15 (25.0)	8 (26.7)	0.864
Melanoma	2 (3.3)	0 (0.0)	0.312
Gastric cancer	8 (13.3)	10 (33.3)	0.025
Hematological n (%)
Anemia	34 (56.7)	4 (15.4)	0.001
Iron deficiency anemia	27 (45.0)	3 (11.5)	0.003
Vitamin B12 deficiency anemia	17 (28.3)	1 (3.9)	0.030
Only iron deficiency anemia	15 (25.0)	2 (7.7)	0.064
Only vitamin B12 deficiency anemia	5 (8.3)	0 (0.0)	0.129
Neuropsychiatric n (%)
Peripheral neuropathy	7 (11.7)	0 (0.0)	0.051
Depression	17 (28.3)	6 (20.0)	0.393
Anxiety disorder	4 (6.7)	3 (10.0)	0.578
Metabolic n (%)
Arterial hypertension	11 (18.3)	8 (26.7)	0.361
Type 2 diabetes mellitus	4 (6.7)	3 (10.0)	0.578
Fatty liver disease	13 (21.7)	3 (10.0)	0.172
Dyslipidemia	18 (30.0)	9 (30.0)	1.000
Gastric neuroendocrine tumors n (%)	7 (11.7)	0 (0.0)	0.051
Other extra-gastrointestinal neoplasm n (%) c	4 (6.7)	2 (6.7)	1.000
Helicobacter pylori (%) d	20 (33.3)	9 (30.0)	0.758
General immune-mediated diseases n (%)
Hypothyroidisme	30 (50.0)	4 (13.3)	0.001
Hyperthyroidism	6 (10.0)	0 (0.0)	0.073
Rheumatoid arthritis	3 (5.0)	1 (3.3)	0.718
Spondyloarthropathy	2 (3.3)	0 (0.0)	0.312
Coeliac Disease	5 (8.3)	0 (0.0)	0.104
Type 1 diabetes mellitus	1 (1.7)	0 (0.0)	0.477
Suprarrenal insufficiency	1 (1.7)	0 (0.0)	0.477
Polymyalgia rheumatica	1 (1.7)	0 (0.0)	0.477

^aStudent t-test for age and chi-square test were used for categorical variables. Significance p<0.05.

^bHistory of current and previous supplementation.

^cCases: 1 breast cancer, 2 thyroid cancer, 1 cervical cancer. Controls: 2 prostate cancer.

^dPresence of H. pylori on gastric biopsy.

^eCases of hypothyroidism secondary to thyroidectomy for thyroid cancer and cases secondary to hyperthyroidism treatment were excluded.

Of patients with a diagnosis of AIG, the vast majority (n=55, 91.7%) had positive serologies, with the majority of these having positive serum PCAs (n=54, 90%), while only a minority had positive IFAs (n=9, 15%), or both (n=8, 13.3%). All patients in the AIG group, including the 5 (8.3%) patients with seronegative AIG, demonstrated histological features of AIG on appropriate gastric biopsies. Corpus-atrophy scores were as follows: absent 4 (6.7%), mild 6 (10%), moderate 17 (28.3%) and intense 33 (55%). All the patients with absent corpus-atrophy were seropositive.

Compared to controls, patients with AIG more often had anemia [56.7% vs. 15.4%; aOR 9.06 (95%CI: 2.61–31.40)] that was attributed to iron deficiency [45.0% vs. 11.5%; aOR 8.50 (95%CI: 2.08–34.80)] or vitamin B12 deficiency [28.3% vs. 3.9%; aOR 10.42 (95%CI: 1.26–86.44)], alone or combined. Iron and vitamin B12 supplementation were reported more often in patients with vs. without AIG. A history of peripheral neuropathy was reported in 11.7% (n=7) of patients with AIG, and no cases in the control group (p=0.051).

Regarding general immune-mediated diseases, patients with AIG had a significantly higher frequency and likelihood of comorbid hypothyroidism [50.0% vs. 13.3%; aOR 8.02 (95%CI: 2.26–28.52)]. There were no statistically significant differences in the frequency or likelihood of other general immune-mediated diseases between cases and controls.

Gastric neuroendocrine tumors (NETs) were detected in 11.7% (n=7) of patients with AIG, compared to no patients in the control group (p=0.051). There were no diagnoses of gastric adenocarcinoma observed in either cases or controls.

Histological presence of H. pylori was common, but there was no significant difference in frequency or likelihood of having positive H. pylori in the gastric biopsy in patients with vs. without AIG [33.3% vs. 30%; aOR 1.18 (95%CI: 0.44–3.16)].

Mucocutaneous manifestations

MC manifestations according to study group are summarized in Table 2 and the frequency of presentation of patients with “pooled immune-mediated diseases” and “pooled mucocutaneous immune-mediated diseases” are summarized in Table 3.

Table 2.

Clinical findings in skin, mucosa and annexes.

Mucocutaneous findings	Autoimmune gastritis (n=60)	Controls (n=30)	p valuea
I. Pigmentary n (%)
Any pigmentary finding	24 (40.0)	10 (33.3)	0.539
Vitiligo	4 (6.7)	0 (0.0)	0.110
II. Alopecia n (%)
Any alopecia	26 (43.3)	10 (33.3)	0.361
Alopecia areata	5 (8.3)	2 (6.7)	0.415
Telogen effluvium	2 (3.3)	2 (6.7)	0.469
Androgenetic alopecia	19 (31.7)	9 (30)	0.872
Lichen planus pilaris	3 (5.0)	1 (3.3)	0.744
III. Sicca n (%)
Sjögren’s syndrome	5 (8.3)	0 (0.0)	0.104
Xerosis	35 (58.3)	10 (33.3)	0.004
Xerostomia	11 (18.3)	1 (3.3)	0.064
Xerophtalmia	18 (30)	2 (6.7)	0.022
IV. Oral cavity n (%)
Any finding in oral cavityb	33 (55.0)	14 (46.7)	0.456
Painful mouth syndrome	3 (5.0)	0 (0.0)	0.213
Recurrent aphthae	5 (8.3)	0 (0.0)	0.104
Angular cheilitis	1 (1.7)	1 (3.3)	0.613
Gums	16 (26.7)	11 (36.7)	0.329
Gingivitis	6 (10.0)	3 (10.0)	1.000
Cobblestone	13 (21.7)	5 (16.7)	0.576
Melanosis	4 (6.7)	3 (10.0)	0.578
Tongue	11 (18.3)	1 (3.3)	0.063
Depapillated tongue	4 (6.7)	0 (0.0)	0.148
Fissured tongue	7 (11.7)	1 (3.3)	0.223
Geographical tongue	1 (1.7)	0 (0.0)	0.477
Other findings in oral cavityc	10 (16.7)	2 (6.7)	0.188
V. Nails n (%)
Any nail finding (non-infectious)d	27 (52.9)	7 (28.0)	0.053
Nail pits	1 (1.7)	0 (0.0)	0.477
Nail fragility	10 (19.6)	0 (0.0)	0.018
Koilonychia	4 (6.7)	0 (0.0)	0.148
Onychoschizia	4 (6.7)	2 (6.7)	1.000
Onychodystrophy	1 (1.7)	1 (3.3)	0.613
Onychorrhexis	10 (16.7)	5 (16.7)	1.000
Trachyonychia	3 (5.0)	0 (0.0)	0.213
Onycholysis	3 (5.0)	1 (3.3)	0.718
Onychomadesis	1 (1.7)	0 (0.0)	0.477
Onychocryptosis	1 (1.7)	0 (0.0)	0.477
Acropachy	1 (1.7)	0 (0.0)	0.477
Nail colour change	4 (6.7)	1 (3.3)	0.515
VI. Other mucocutaneous immune-mediated diseases n (%)
Urticaria	1 (1.7)	0 (0.0)	0.477
Lupus	1 (1.7)	0 (0.0)	0.477
Dermatomyositis	0 (0.0)	0 (0.0)	1.000
Morphea	1 (1.7)	0 (0.0)	0.477
Lichen sclerosus	1 (1.7)	0 (0.0)	0.477
Vasculitis	0 (0.0)	0 (0.0)	1.000
Panniculitis	0 (0.0)	0 (0.0)	1.000
Psoriasis	2 (3.2)	2 (6.7)	0.469
Hidradenitis suppurativa	2 (3.3)	0 (0.0)	0.312
VII. Infectious diseases n (%)
Fungal infectione	15 (25.0)	9 (30.0)	0.613
Onychomycosis	10 (16.7)	7 (23.3)	0.446
Tinea pedis	5 (8.3)	4 (13.3)	0.456
Bacterial infection	2 (3.3)	2 (6.7)	0.469
Viral infectionf	0 (0.0)	2 (6.7)	0.043
VIII. General cutaneous findings n (%)
Rosacea	10 (16.7)	5 (16.7)	1.000
Acne	5 (8.3)	2 (6.7)	0.781
Seborrheic dermatitis	8 (13.3)	2 (6.7)	0.343
Dyshidrosis	3 (5.0)	2 (6.7)	0.745
Chronic lichen simplex	3 (5.0)	1 (3.3)	0.718
Other dermatitis	8 (13.3)	7 (23.3)	0.230
Pilar keratosis	0 (0.0)	1 (3.3)	0.155
Multiple acrochordons	8 (13.3)	4 (13.3)	1.000
Acanthosis nigricans	3 (5.0)	5 (16.7)	0.067
Multiple lentigos	13 (21.7)	8 (26.7)	0.597
Multiple seborrheic keratoses	9 (15.0)	4 (13.3)	0.832
Multiple nevi	3 (5.0)	2 (6.7)	0.745
Atypical nevus (clinical suspicion)	7 (11.7)	4 (13.3)	0.820
Malignant cutaneous neoplasmg	3 (5.0)	0 (0.0)	0.213

^aChi-square test was used for categorical variables. Significant associations confirmed by a multivariable logistic regression modeling adjusted for sex, age, H. pylori, tobacco use and alcohol consumption. Significance p<0.05.

^bIncludes “Xerostomia”, which was also analyzed separately in “Sicca”.

^cIncludes petechiae on the palate, vitiligo achromy, 1 case with hypertrophic papillae.

^dOnychomycosis and acrylic or enameled nails that made its examination impossible were excluded.

^e“Fungal infection” also includes pityriasis versicolor and balanitis.

^f“Viral infection” includes 2 warts in the control group.

^gIncludes 3 cases of basal cell carcinoma in the AIG group.

General mucocutaneous manifestations

Compared to controls, patients with AIG more often presented with sicca clinical manifestations, including xerosis [58.3% vs. 33.3%; aOR 5.05 (95%CI: 1.68–15.20)], xerostomia [18.3% vs. 3.3%; aOR 8.16 (95%CI: 0.89–75.13)] and xerophthalmia [30% vs. 6.7%; aOR 6.41 (95%CI: 1.32–31.22)]. There were also differences in oral cavity findings. Pooled lingual findings, such as depapillated, fissured and geographic tongue, were more frequent in patients with vs. without AIG [18.3% vs. 3.3%; aOR 7.77 (95%CI: 0.90–67.44)] (Figure 1). Also, painful mouth syndrome and recurrent aphthae were only detected in cases, with a frequency of 5% (n=3) and 8.3% (n=5) respectively, but not in controls. Non-infectious nail changes were also more frequent in patients with AIG compared to controls [52.9% vs. 28%; aOR 2.85 (95%CI: 0.99–8.27)]; nail fragility was the most frequent finding (19.6% vs. 0.0%; p=0.018).

Figure 1. General mucocutaneous findings:

Geographic tongue (a), hypertrophic papillae (b), fissured/hairy black tongue (c) and scrotal tongue (d).

Pooled immune-mediated diseases and mucocutaneous immune-mediated diseases

Pooled immune-mediated diseases (as defined above) were significantly more frequently observed among patients with AIG compared to controls, with a 11.7-fold and 5.6-fold higher likelihood, irrespective of whether immune-thyroid diseases were included [70.0% vs. 26.7%; aOR 11.67 (95%CI: 3.49–39.05)] or excluded [38.3% vs. 13.3%; aOR 5.55 (95%CI: 1.53–20.17)]. Similarly, pooled MC immune-mediated diseases were more frequently observed in patients with vs. without AIG, with a 12-fold and 4.8-fold higher likelihood, irrespective of whether immune-thyroid diseases were included [66.7% vs. 23.3%; aOR 12.02 (95%CI: 3.51–41.13)] or excluded [30.0% vs. 10.0%; aOR 4.75 (95%CI: 1.18–19.16)].

We also evaluated the association between PCAs or IFAs serologies and immune manifestations in patients with AIG. Patients seropositive vs. seronegative for IFAs more often had concomitant pooled immune-mediated diseases (100% vs. 62.3%; p=0.025) and concomitant pooled MC immune-mediated diseases (100% vs. 58.5%; p=0.016), both including immune-thyroid diseases. By contrast, there was no significant difference in the frequency of these pooled immune-mediated conditions (MC and non-MC) in patients with AIG seropositive vs. seronegative for PCAs or both serologies (data not shown).

The most common MC immune-mediated conditions were alopecia areata 8.3% (n=5), Sjögren’s syndrome 8.3% (n=5), coeliac disease 8.3% (n=5), and vitiligo 6.7% (n=4), (Figure 2). By contrast, among patients without AIG, only 6.7% (n=2) of patients were diagnosed with alopecia areata. The exploratory hypothesized conditions (vitiligo, coeliac disease, alopecia areata or Sjögren’s syndrome) were diagnosed in 25.0% (n=15) of patients with AIG compared to only 6.7% (n=2) of controls; this corresponded to an aOR of 6.82 (95%CI: 1.27–36.72).

Figure 2. Vitiligo:

Case 1: 71-years old man. History of acrofacial (a), mucosal (b) and perianal (c) vitiligo, psoriasis vulgar, hypothyroidism, fissured tongue, nail fragility and koilonychia. Additionally, previous history of vitamin B12 deficiency and depression, anti-thyroid antibodies (+), ANA (+) 1/80 and elevated gastrin levels (138 pg/mL).

Case 2: 58-years old woman. History of periocular (d) (e) and perianal (f) vitiligo, hypothyroidism, anemia due to iron deficiency and depression. Wood’s Lamp was used to highlight periocular vitiligo (e).

Anecdotally, the most remarkable case of polyautoimmunity in the AIG group was a 45-year-old man diagnosed with alopecia areata universalis, Sjögren’s syndrome, vitiligo and genital lichen sclerosus (Figure 3). Additionally, he had a previous history of serum vitamin B12 deficiency (serum B12<100 pg/mL), peripheral neuropathy and burning mouth syndrome. On serological analysis, he demonstrated positivity for PCAs, anti-thyroid antibodies, antinuclear antibodies (ANA) (1:80 titer) and elevated serum gastrin level (331 pg/mL; reference: 13–115 pg/mL).

Figure 3. Polyautoimmunity:

Case 3: 45-years old man. History of alopecia areata universalis (a), trachyonychia (b), periodontal disease secondary to Sjögren’s syndrome (c), vitiligo and genital lichen sclerosus (d). Additionally, previous history of vitamin B12<100 pg/mL, peripheral neuropathy, burning mouth syndrome, anti-thyroid antibodies (+), ANA (+) 1/80 and elevated gastrin levels (331 pg/mL).

Laboratory profile

Complete serum biochemical evaluation and autoimmune serology according to case vs. control status are summarized in Table 4 and Table 5.

Table 4.

Laboratory profile.

	Autoimmune gastritis (n=60)	Controls (n=30)	p value
Laboratory test results, median (IQR) a
Hematocrit (%)	38.9 (37.2–42.2)	40.4 (38.9–41.4)	0.172
Hemoglobin (g/dL)	13.1 (12.5–14.2)	13.7 (13.1–14.2)	0.152
MCV (f/L)	88.9 (85.0–93.0)	88.3 (86.0–93.1)	0.859
Leukocytes (x 10^3/μL)	5.9 (4.9–7.0)	6.4 (5.3–7.6)	0.185
Platelets (x 10^3/μL)	267.0 (198.5–301.5)	274.0 (215.0–320.0)	0.378
ESR (mm/hr)	8.0 (5.0–15.5)	9.0 (5.0–13.0)	0.976
Vitamin D (ng/mL)	20.2 (15.6–26.6)	16.2 (14.0–22.6)	0.144
Vitamin B12 (pg/mL)	245.0 (140.0–440.0)	439.0 (324.0–624.0)	0.005
Ferritin (ng/mL)	37.4 (12.6–115.5)	37.8 (22.1–88.6)	0.702
Gastrin (pg/mL)	395.0 (103.0–902.0)	102.0 (47.0–137.0)	0.047
Proportion of alteration in laboratory tests, n (%) b
Hb <12 Female / Hb <13 Male (g/dL)	15 (25.0)	3 (11.5)	0.159
MCV <80 (f/L)	7 (11.7)	0 (0.0)	0.081
MCV >100 (f/L)	7 (11.7)	0 (0.0)	0.081
Vitamin B12 < 200 (pg/mL)	26 (44.1)	1 (5.6)	0.003
Vitamin D < 20 (ng/mL)	26 (46.4)	10 (62.5)	0.257
Vitamin D < 30 (ng/mL)	46 (82.1)	15 (93.8)	0.255
Ferritin < 30 (ng/mL)	25 (41.7)	6 (37.5)	0.763
Iron deficiencyc	25 (42.4)	6 (46.2)	0.071

Significant associations confirmed by multivariable logistic regression modeling adjusted for sex, age, H. pylori, tobacco and alcohol consumption. Significance p<0.05.

^aDifferences calculated by chi-square test.

^bDifferences calculated by Mann-Whitney test.

^cFerritin <30 ng/mL or transferrin saturation <16%.

MCV: mean corpuscular value; ESR: erythrocyte sedimentation rate; Hb: hemoglobin.

Table 5.

Autoimmune serologic profile of patients with AIG.

Serologic evaluation n (%)	Autoimmune gastritis (n=60)
Anti-parietal cell antibodies (PCAs)	54 (90.0)
Anti-intrinsic factor antibodies (IFAs)	9 (15.0)
PCAs or IFASs	55 (91.7)
PCAs and IFAs	8 (13.3)
Antinuclear antibodies (ANA)	15 (25.0)
Anti-Thyroglobulin antibodies (Tg)	32 (53.3)
Anti-Thyroid peroxidase antibodies (TPO)	34 (56.7)
Tg or TPO	39 (65.0)
Tg and TPO	27 (45.0)
Anti-Endomysial antibodies (EMA)	2 (3.3)
Anti-tissue Transglutaminase antibodies (tTG)	1 (1.7)
EMA or tTG	2 (3.3)
EMA and tTG	1 (1.7)
IGA<70 mg/dL	3 (5.0)

PCAs, anti-gastric parietal cells antibodies; IFAs, anti-intrinsic factor antibodies; ANA, antinuclear antibodies; TG, anti-thyroglobulin antibodies; TPO, anti-thyroid peroxidase antibodies; EMA, anti-endomysial antibodies; tTG, anti-tissue transglutaminase antibodies; IGA, immunoglobulin A.

As expected, patients with AIG had lower median serum vitamin B12 levels compared to controls (245 vs. 439 pg/mL; p=0.005), with vitamin B12 deficiency (<200 pg/mL) more frequently observed patients with vs. without AIG (44.1% vs. 5.6%; p=0.003). A diagnosis of AIG was associated with a 25-fold higher likelihood of vitamin B12 deficiency (aOR 25.13 95%CI: 3.08–204.91). Also, serum gastrin levels were significantly higher in patients with AIG compared to controls (395 vs. 102 pg/mL; p=0.047), despite significantly higher proton pump inhibitors usage among controls (43.3% vs. 66.7%; p=0.037). No significant differences were observed in hematocrit, vitamin D, serum ferritin or transferrin saturation between cases and controls.

Serum anti-nuclear antibodies (ANA) were found in 25% (n=15), anti-thyroid peroxidase (TPO) or anti-thyroglobulin (TG) antibodies in 65% (n=39) and anti-tissue transglutaminase (tTG) or anti-endomysial (EMA) antibodies in 3.3% (n=2) of patients with AIG.

DISCUSSION

To our knowledge, this is the first comprehensive, comparative study to analyze MC manifestations in patients with vs. without AIG, as well as quantify the magnitude of association between AIG and specific MC manifestations. Here we report for the first time a significant consistently higher frequency of immune- and nutritional deficiency-related MC manifestations in a well-characterized group of patients with AIG compared to controls, who were histologically confirmed to not have AIG or any chronic atrophic gastritis (OLGA 0). The main immune diseases with MC involvement associated with AIG were vitiligo, Sjögren’s syndrome and alopecia areata. In addition, xerosis, lingual and nail disorders which are typically associated with nutritional deficiencies, were common MC findings in AIG, observed in more than half of these patients.

It is well-documented that AIG is associated with an approximate 3–5 times higher likelihood of having another autoimmune diseases, which may have MC involvement.[23, 24] In a retrospective study that evaluated polyautoimmunity in a cohort of 320 patients with AIG, 53% had another concomitant autoimmune disease, 42% with only one and 11% with more than one; however, they did not specifically evaluate MC manifestations.[25] There are reports of MC involvement in immune-mediated diseases associated with AIG as diabetes mellitus type-1,[26] Sjögren’s syndrome,[27] vitiligo,[28, 29] alopecia areata,[30] and chronic urticaria.[31] In AIG, small reports have documented mucosal manifestations including atrophic glossitis, burning mouth syndrome and erosive oral lichen planus.[32] Concordantly with our findings, when pooling all the immune-mediated diseases with or without previous documentation of MC involvement in the literature, we still identified a strong and significant association with AIG, which was even more pronounced when including immune-thyroid diseases.

Each of the MC immune-mediated diseases which we evaluated were more frequent in patients with AIG vs. controls, except for psoriasis. Indeed, the majority of MC findings were not even identified among the patients without AIG (controls). That said, due to small per strata numbers, we did not identify a significant difference in the risk of individual MC diseases such as vitiligo, Sjögren’s syndrome and alopecia areata among patients with AIG. Of note, although celiac disease was included in the investigational hypotheses,[33] no cases of dermatitis herpetiformis were observed. This may be due to the fact that all patients with celiac disease were on a gluten-free diet. Notably, other recognizable MC manifestations were still diagnosed among these patients with celiac disease, namely depapillated and fissured tongue, and recurrent oral aphthae. AIG was associated with xerosis, xerostomia and xerophthalmia. Given these conditions are classic for Sjogren’s syndrome, it is possible these simply represent indolent Sjögren’s syndrome.

The role of H. pylori, particularly the more virulent strains, as a trigger of autoimmune diseases has been hypothesized.[34] To evaluate this possible confounding factor we also adjusted by the presence of H. pylori on gastric biopsies; indeed, the positive association of AIG with MC immune-mediated diseases persisted in the fully adjusted model.

Regarding the MC manifestations associated with nutritional deficiencies, over half of patients with AIG had some positive finding in the oral cavity, primarily on the tongue (e.g. depapillated tongue and fissured tongue). Also, painful mouth syndrome and recurrent aphthae were more frequent in AIG. Similarly, more than half of the cases had some non-infectious nail changes, highlighting nail fragility by history or clinical examination. These observations in our cohort are consistent with the iron and vitamin B12 deficiency in patients with AIG, as reported in the literature.[35–37] An important consideration is that deficiency findings could be masked, since at the time of the physical examination most of the patients with AIG reported previous or current supplementation. Immunological MC manifestations were not adjusted for nutritional deficiencies because it might lead to a statistical overcorrection, considering that iron and vitamin B12 deficiency are part of the natural history of AIG and have different pathophysiological mechanisms compared with the other immune diseases.

There were no significant differences in general and cutaneous malignant neoplasms between cases and controls, although some immune-mediated diseases are associated with cancer risk. Three basal cell carcinomas were diagnosed in patients with AIG compared with no cases of cutaneous malignancies among controls. There was no statistically significant difference between multiple lentigos, nevi and seborrheic keratoses. Gastric neuroendocrine tumors (NETs) were detected in cases but not controls, which is expected given the established strong association between AIG and gastric NETs.[38–41] Further, it should be noted that our study was not designed to evaluate the natural history of MC disorders in AIG and we cannot comment on whether the presence or absence of specific MC disorders is associated with AIG-related clinical outcomes, including gastric malignancy.

Finally, we characterized the serological profile of the AIG group because we expected that some MC manifestations might have autoimmune underpinnings. Indeed, ANA was positive in about 25% of patients with AIG; this frequency is similar to a Chinese cohort of 320 patients with AIG, 29.2% of whom demonstrated ANA positivity.[42] Although the role of ANA in AIG is unclear, its presence supports the association with other immune diseases. Furthermore, although IFAs are less often positive in AIG compared to PCAs, we found that all patients with AIG who were also IFAs seropositive had both MC and non-MC immune diseases. Indeed, IFAs are more specific but less sensitive in AIG and appear later in the course, and might underlie this observation. This finding warrants confirmation in other cohorts.

This study has several strengths. It is a novel interdisciplinary prospective case-control study of a well characterized group of patients with vs. without histologically confirmed AIG, with complete dermatological evaluation performed by three expert dermatologists; comprehensive serum hematological, autoimmune and nutritional evaluation; and standardized EGD with mapping gastric biopsies assessed by two expert pathologists.[43] Moreover, controls were histologically confirmed to have OLGA 0 and also underwent the same comprehensive dermatological evaluation. Our study is not without limitations, however. One main limitation is that some comparisons might not be statistically significant due to insufficient power; although, we a priori calculated the sample size needed for the primary outcome. Even though we adjusted by the presence of H. pylori on gastric biopsies, we do not have data on previous eradication treatments and outcomes. Study design limitations include: the non-blinded dermatologists (due to an inherent study design barrier), and that determination of case vs. controls status occurred as a 2-stage protocol due to logistical and epidemiological considerations. As AIG is an infrequent disease, first the cases were collected and evaluated, and then a control group was included. This is a single center study so further multi-centric studies could be performed to validate our findings and enhance generalizability, particularly if these findings are validated in other, more racially and ethnically diverse populations. Future studies are needed to clarify whether MC manifestations in AIG are associated with altered patient-important outcomes.

In conclusion, this is the first case-control study that specifically evaluates MC manifestations in patients with AIG compared to histologically confirmed controls. AIG is more frequently associated with both immune- and nutritional deficiency-related MC manifestations. Based on our study, recognizing MC manifestations associated with AIG is an important aspect of diagnosis, disease characterization and management, particularly with respect to correction of underlying nutritional deficiencies if present. Indeed, our findings suggest that if certain MC findings are seen, particularly in the presence of PCAs or IFAs, formal workup for AIG with EGD and biopsies is reasonable. Future studies are indicated to define whether certain MC manifestations in AIG are associated with prognosis or altered natural disease course.

Significance of this study.

What is known?

• Autoimmune atrophic gastritis (AIG) affects the gastric parietal mucosa.

• AIG is associated with nutritional deficiencies, autoimmune diseases and an increased risk of gastric neuroendocrine tumors.

What is new here?

• This is the first study that comprehensively delineates mucocutaneous manifestations in patients with vs. without AIG based on histological confirmation.

• AIG is significantly more frequently associated with both immune-mediated diseases with mucocutaneous involvement and mucocutaneous manifestations of nutritional deficiencies.

• Identifying the mucocutaneous manifestations associated with AIG could improve its diagnosis and management, as well as the earlier assessment and correction of nutritional deficiencies.

• If these mucocutaneous findings are seen, along with consistent lab derangements, such as iron or vitamin B12 deficiencies, anti-parietal cell antibodies or anti-intrinsic factor antibodies, formal workup for AIG including esophagogastroduodenoscopy with mapping biopsies is reasonable.

Data availability statement:

All data relevant to the study are included in the article. For additional information please contact to the corresponding author.