The Significance and Clinical Outcome of Lymphocytic Duodenosis in Children: Mayo Clinic Experience and Systematic Review

Abstract

Background: 

Lymphocytic duodenosis (LD) defined as increased intraepithelial lymphocytes >25 intraepithelial lymphocytes (IELs) per 100 epithelial cells with normal villous architecture is associated with many gastrointestinal (GI) disorders. We aim to assess the rate and outcome of LD in children and perform a systematic review.

Method: 

We reviewed all children (<18 years) who underwent esophagogastroduodenoscopy (EGD) with duodenal biopsy between January 2000 and June 2019 to identify LD cases and control group. Demographics, clinical, and pathologic information were reviewed and recorded. A systematic review including our findings was performed.

Results: 

During the study period 12,744 children underwent an EGD with biopsies. Of those, we identified 426 children with LD (3%) and 474 controls. The median age in years was 10.7 and 12.6 and there were 254 (60%) and 278 (59%) girls in the LD and control group, respectively. The most common presenting symptoms in both groups were abdominal pain (52%), gastroesophageal acid reflux disease (18%), diarrhea (16%), and vomiting (12%). Diarrhea (21% vs 12%, P < 0.001) and constipation (2% vs 0.4%, P = 0.021) were statistically different between the LD and control group, respectively. Median follow-up (range) is 3.6 (0.0, 190.9) and 3.1 (0.0, 194.2) in the LD and control group, respectively. CD (5% vs 0%, P < 0.001), Crohn disease (9% vs 3%, P = 0.003) and Helicobacter pylori gastritis (3% vs 1%, P = 0.021) were more common in the LD group.

Conclusions: 

The Rate of LD in children is similar to reported rate in adults. In the absence of Crohn disease, CD or H. Pylori, LD seems to be a benign and transient histologic finding in children.

The authors report no conflicts of interest.

What Is Known

• Lymphocytic duodenosis is relatively common finding on small bowel biopsy.
• Lymphocytic duodenosis corresponds with celiac disease histologic changes.

What Is New

• Rate of Lymphocytic duodenosis in children in similar to adults.
• Lymphocytic duodenosis was more common in children presenting with diarrhea.
• Even though Crohn, celiac and Helicobacter pylori diseases were more common in children with lymphocytic duodenosis, most cases ended up with functional gastrointestinal disorder.

The small bowel villous surface is composed mostly of absorptive epithelial cells (ECs) with occasional interspersed intraepithelial lymphocytes (IELs). Those IELs are an essential part of the small bowel defense and surveillance system that responds to the stress of antigenic triggers ^(1–4). The clinical significance of identifying an increase in the number of IELs (>25/100 ECs) with intact villous architecture, defined as lymphocytic duodenosis (LD) is still unclear to pathologists and clinicians. With the evolution of pediatric gastroenterology as a specialty, the number of gastrointestinal (GI) procedures in children have increased over the last decades ⁽⁵⁾, hence more histologic changes without clear clinical significance like mild chronic gastritis and LD are reported ^(5–8). LD is relatively a common histologic finding on duodenal biopsies of children undergoing EGD because of GI symptoms. Previous studies that were mostly adult studies suggested that LD can be found in up to 3.8% of duodenal biopsies ⁽⁹⁾. LD also corresponds to grade1 of Marsh classification that describes the early histologic changes of celiac disease (CD), hence the presence of LD may result in adding to pathology reports the following comment “LD can be seen in symptomatic, latent, or partially treated CD, dermatitis herpetiformis, and First-degree relatives of patients with CD” ⁽¹⁰⁾. On the basis of that, previously published studies have focused on the risk of CD in patients with LD. It has been suggested that up to 10% of adult LD cases end up with CD diagnosis ⁽⁸⁾. It is important to note that LD has also been found to be associated with other disorders like inflammatory bowel disease (IBD) (Crohn disease and ulcerative colitis), cow's milk protein intolerance, GI infections, Helicobacter infection, and the use of medications like nonsteroidal anti-inflammatory drugs (NSAID) ^(8,11–17).

Different approaches have been tried to differentiate between LD cases that belong to celiac spectrum, other GI disorders, and the benign transient normal immune reaction. Those approaches may include further blood and stool studies to exclude other immunologic and inflammatory disorders. Other approaches included trial of restrictive dietary changes like gluten-free diet (GFD) with repeat SBB or following symptom response to dietary changes ^(13,14). It is not clear if LD is being reported more frequently over time and if the anxiety about identifying it is justified. Despite the relatively low rate of significant GI pathology between LD cases, there is a danger of over diagnosis or even unnecessary testing solely based on the presence LD. One major limitation to the available literature and previous studies on LD is the paucity of data about its frequency and clinical significance in children. Our aim is to assess the rate of LD in children and describe their clinical outcome along with performing systematic review of the literature.

METHODS

Mayo Clinic Retrospective Study

We performed a retrospective study using the Mayo Clinic electronic medical records (EMR) and pathology database between January 1, 2000 and 30 June, 2019. We identified all children (<18 years) with GI symptoms who underwent their first diagnostic esophagogastroduodenoscopy (EGD), children with previous GI diagnosis were excluded. We used the keywords normal villi, preserved villous architecture, increased lymphocytes, and epithelial lymphocytosis to identify the study subjects. Children with LD were included as LD group. Of the children with GI symptoms who underwent EGD during the study period, we used a simple random sample to select a set of children (control group) who also underwent their first diagnostic EGD with normal SBB to serve as a comparison group. Demographics, clinical, and pathologic information were collected, including age at the time of EGD, sex, proton pump inhibitors (PPI) use, recent infections, and antibiotic use within 3 months of EGD. In our cohort, CD diagnosis was considered in the children if they had positive celiac serologic markers along with GI symptoms that improved on GFD. Celiac serologic markers that were done included serum immunoglobulin A (IgA), anti-tissue transglutaminase immunoglobulin A (TTG IgA), and endomysial immunoglobulin A antibody (EMA). If serum IgA was low, then immunoglobulin G (IgG)-based serologic markers like TTG IgG, deamidated gliadin antibodies IgG were performed. The presence of celiac predisposing HLA types was supportive of the CD diagnosis, but not necessary as it was not performed on all children. The final diagnoses were recorded and compared between the 2 groups. Children with a previous GI diagnosis were excluded. This study was approved by the Mayo Clinic institutional review board (IRB).

Statistical Analysis

Continuous and categorical measures were summarized within each group. The median, minimum, and maximum were used to summarize numerical variables including age and laboratory results while counts and percentages were used for categorical variables, such as presenting symptoms and PPI use. Associations between clinical factors and LD were assessed using the Kruskal-Wallis and Chi-square tests for continuous and categorical variables, respectively. A type I error rate of 0.05 was used for all analyses, and no adjustment was made for multiple comparisons. SAS (v9.4M5, Cary, NC) was used for all analysis.

Systematic Review and Meta-analysis

The systematic review of the literature was reported according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines.

Eligibility Criteria

We included studies that looked for the frequency of LD in patients undergoing EGD and the characteristics of these patients. English full-text articles were included, whereas case reports were excluded.

Data Sources and Search Strategies

A comprehensive search of several databases from inception to November 1, 2019, excluding animal studies, was conducted. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process ānd Other Non-Indexed Citations and Daily, Ovid Embase, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. The search strategy was designed and conducted by an experienced librarian with input from the study's principal investigator. Controlled vocabulary supplemented with keywords was used to search for studies of interest. The full search strategy is available in Supplemental data 2, https://links.lww.com/MPG/B969.

Study Selection and Data Extraction

Two independent reviewers identified the articles for inclusion. These 2 reviewers independently extracted the data from the included studies. Any conflict between the reviewers was solved by consensus.

Risk of Bias

The risk of bias for included studies was assessed by using a modified Newcastle Ottawa scale for risk of bias assessment for observational studies (Supplemental data 3, https://links.lww.com/MPG/B970).

Analysis

We used random-effect model meta-analyses to estimate pooled ratios and 95% confidence intervals (CIs). We estimated inconsistency between trials not because of chance using the I² statistic.

RESULTS

Mayo Clinic Retrospective Study

During the study period, 12,744 children underwent diagnostic EGD with biopsies. Of those 6468 (51%) were girls and 6276 (49%) boys. We identified 510 (4%) children with LD on their duodenal biopsies. Of those 426 (3%) children had LD identified on their first diagnostic EGD and were included in this study as LD group after excluding 84 children with known GI diagnosis including CD, IBD, EOE, and GERD. Four hundred seventy-four (4%) children were randomly selected as control group. The median age (range) at the time of EGD in the LD group was 10.7 (0.2, 18.0) years with 254 (60%) girls. In the control group, the median age (range) at the time of EGD was 12.6 (0.0, 18.0) years with 278 (59%) girls. The most common presenting GI symptoms at the time of EGD were mostly similar in both groups and many children had more than 1 presenting symptom. Diarrhea and constipation were more common in the LD group (Table 1). When comparing the previous exposure (within 3 months from EGD) to PPI and antibiotics between the 2 groups, we found that PPI use (41% vs 28%, P < 0.001) and antibiotic exposure (20% vs 11%, P < 0.001) were more common in the LD group. History of recent infections (viral or bacterial) was similar at 26% and 28% in the LD and control groups, respectively (P = 0.365) (Table 1).

TABLE 1 - Demographics, presenting symptoms, and diagnoses details

	LD group N = 426	Control group N = 474	P value
Age (years) at EGD, median (range)	10.7 (0.2, 18.0)	12.6 (0.0, 18.0)	<0.001∗
Female gender, n (%)	254 (60%)	278 (59%)	0.767†
Presenting symptoms, n (%)
Abdominal pain	228 (54%)	244 (51%)	0.540†
Diarrhea	88 (21%)	56 (12%)	<0.001†
Vomiting	41 (10%)	64 (14%)	0.070†
Reflux	77 (18%)	85 (18%)	0.956†
Constipation	9 (2%)	2 (0%)	0.021†
Failure to thrive	11 (3%)	7 (1%)	0.237†
Other symptom	50 (12%)	70 (15%)	0.182†
Prior exposures, n (%)
Prior PPI use	176 (41%)	131 (28%)	<0.001†
Prior antibiotic use	86 (20%)	51 (11%)	<0.001†
Recent illness	109 (26%)	134 (28%)	0.365†
Follow-up time (months) since EGD, median (range)	3.6 (0.0, 190.9)	3.1 (0.0, 194.2)	0.790∗

EGD = esophagogastroduodenoscopy; LD = lymphocytic duodenosis; PPI = proton pump inhibitors.

^∗Kruskal-Wallis test.

^†Chi-square test.

Celiac serologic markers were ordered more often in the LD group 333/426 (78%) compared with 216/474 (46%) in the control group (P < 0.001). Celiac serology tests are reported in Table 2. In the LD group, celiac serologic markers were more commonly ordered after identifying LD in 134/333 (55%) of the cases. Of the 333 children with LD who had celiac serologic testing, 21 children had positive celiac serology (19 TTG IgA, 1 IgA deficient with positive TTG IgG and 1 isolated TTG IgG with normal serum IgA). Of those, only 20/426 (5%) met the CD diagnostic criteria (1 child had isolated positive TTG IgG with normal IgA and negative TTG IgA that did not respond to GFD). In the control group, celiac markers were ordered in 216 (46%) and were normal except for 1 child who had mild positive TTG IgA (<3 upper normal limit) and, negative EMA that normalized on regular diet.

TABLE 2 - Celiac serologic test results

	LD group N = 426	Control group N = 474	P value
Celiac serology performed	333 (78%)	216 (46%)	<0.001∗
IgA abnormal	14/333 (4%)	4/216 (2%)	0.122∗
TTG IgA abnormal	19/327 (6%)	1/208 (0%)	0.002∗
TTG IgG abnormal	16/81 (20%)	1/42 (2%)	0.008∗
Positive celiac serology†	21/338 (6%)	1/209 (0%)	0.001∗
Endomysial antibody positive	15/48 (31%)	0/16 (13%)	0.011∗
DGP IgG abnormal	5/32 (16%)	0/16 (0%)	0.095∗
Celiac genetic test completed	142 (33%)	23 (5%)	<0.001∗
Genetic pairs present	71/142 (50%)	18/23 (78%)	0.012∗

LD = lymphocytic duodenosis; TTG IgA = transglutaminase immunoglobulin A.

^∗Chi-square test.

^†Defined as abnormal TTG IgA or abnormal TTG IgG for normal or deficient IgA, respectively.

Celiac HLA testing was performed in 142 (33%) of the children in the LD group and 23 (5%) in the control group (P < 0.001). HLA testing was positive in 71/142 (50%) and 18/23 (78%) tested children in the LD and control group, respectively. In the LD group, HLA testing was more commonly ordered after identifying LD in 116/142 (82%). In the control group, HLA testing was performed because of family history of CD, persistent symptoms including abdominal pain, diarrhea, and weight loss.

In the nonceliac LD cases, GFD was tired in 36/406 (9%) children despite negative celiac serologic markers. These children were treated with GFD along with other symptomatic management at the same time, so it was hard to assess their response to GFD. Only 6/36 reported improvement of symptoms on average after 2 months on GFD. In the LD group, repeat EGD was done in 92/426 (22%) children (89 nonceliac and 3 CD). Repeat EGD was done within a median (range) of 11.8 (0.6, 146.0) months after their first EGD. Indications for repeating the EGD were persistent symptoms in 57 (including 3 CD) or follow-up on their GI diagnoses in 35 children (13 Crohn disease, 5 ulcerative colitis, 8 EOE, and 9 GERD). In the group of nonceliac 54 children, who underwent EGD due persistent symptoms, repeat EGD showed LD resolution in 45/54 (83%) with a median (range) 10.6 (1.4–146.0) months between the 2 EGDs. Of those only 8/54 (15%) were on GFD at the time of repeat EGD. Only 2 children in the control group with a family history of CD, both with negative celiac serology, were tried on GFD for management of abdominal pain and behavioral concerns; only the child with abdominal pain reported some benefit from being on GFD.

The median (range) of follow-up time was 3.6 (0.0–190.9) months for the LD group compared with 3.1 (0.0–194.2) months for controls (P = 0.790). Final diagnosis data was available for 387/426 (91%) and 383/474 (81%) in the LD and control group, respectively. Among those with a final diagnosis, the LD group had a median (range) follow-up time of 4.4 (0.0, 190.9) months whereas the control group had 6.9 (0.0, 194.2) months in the control group (P = 0.031). On the basis of follow-up data, the most common diagnoses in both groups were functional GI disorders like chronic abdominal pain, irritable bowel syndrome, and constipation. The prevalence of functional GI disorders was similar between the LD group (41%) and the control group (39%, P = 0.444). When compared with the control group, Crohn disease (33, 9% vs 13, 3%, P = 0.003), CD (20, 5% vs 0, 0%, P < 0.001), and H pylori gastritis (10, 3% vs 2, 1%, P = 0.021) were more common in the LD group (Table 3).

TABLE 3 - Outcome details among children with available follow-up data

	LD Group N = 387	Control Group N = 383	P-value
Final Diagnosis Details†, n (%)
Celiac Disease	20 (5%)	0 (0%)	<0.001∗
Crohn's Disease	33 (9%)	13 (3%)	0.003∗
Ulcerative Colitis	13 (3%)	12 (3%)	0.860∗
H. pylori Gastritis	10 (3%)	2 (1%)	0.021∗
EOE	20 (5%)	23 (6%)	0.613∗
GERD	57 (15%)	61 (16%)	0.664∗
Fructose Malabsorption	13 (3%)	23 (6%)	0.082∗
Small Bowell Bacterial  Overgrowth	3 (1%)	18 (5%)	0.001∗
IBS	12 (3%)	28 (7%)	0.008∗
Rumination Syndrome	0 (0%)	8 (2%)	0.004∗
Lactose Intolerance	21 (5%)	39 (10%)	0.014∗

EoE = eosinophilic esophagitis, GERD = gastroesophageal reflux disease, IBD = inflammatory bowel disease, LD = lymphocytic duodenosis.

^∗Chi-square test.

Systematic Review and Meta-analysis

The literature search found 285 potential studies, after the title and abstract screening; 17 studies were identified for full-text screening. Six studies were included in the systematic review (Supplemental 1, https://links.lww.com/MPG/B968). Three of these studies were conducted at Mayo Clinic in the USA, 2 in the UK, and 1 in Ireland. Four of the studies were in adults only, whereas 1 had mixed adults and children, and 1 study included children only. There were 2 studies (Aziz 2010 and Aziz 2015) that included overlapping patients, so we included the most recent publication in this systematic review. The studies had low risk of bias based on the Newcastle–Ottawa scale (Supplemental 4, https://links.lww.com/MPG/B971).

We only could do a meta-analysis for 3 of the adult studies (Mahadeva, 2002, Shmidt, 2014, and Parihar, 2017) to estimate 4% (95% CI: 0.04–0.04) rate of LD of from all EGDs in adults. Rate of LD per population shown in Figure 1. We could not do a meta-analysis of children studies as there are overlapping patients between our study and (Shmidt, 2014). CD was diagnosed in 6% (95% CI: 5%–7%) of adult patients with LD from a meta-analysis of 2 studies (Shmidt, 2014, Parihar, 2017). Our study showed that CD presented in 5% of children with LD. IBD rate of 8% (95% CI: 7%–10%) in adults with LD from a meta-analysis of 2 studies (Shmidt, 2014, Parihar, 2017). Our study showed IBD rate of 12% in children with LD. One study (Kakar, 2003) showed IBD rate of 12% of mixed LD population of children and adults.

FIGURE 1: Rate of lymphocytic duodenosis per population.

DISCUSSION

To our knowledge, this is the largest pediatric study to look at the rate and clinical outcome of LD in children. We looked at all children who underwent EGD over 19 years at the Mayo Clinic. In this cohort, we found that LD was reported in 3% of children who underwent EGD. This is similar to the rate of 3.8% reported by Walker et al in healthy adults with negative celiac serologic markers and lower than the rate of 9% that was reported in previous adult studies ^(8,18). In our cohort Crohn disease, CD and H pylori were more frequently diagnosed in the LD group when compared with control at a rate of 9%, 5%, and 3%, respectively. All children with CD diagnoses had positive celiac serologic markers, whereas the children with IBD had both suggestive symptoms and laboratory testing showing active inflammatory state. Even though a previous study suggested that LD may be clustered toward the tip of villi in the patients with celiac spectrum disorder, there are no clear histological features that can predict which LD cases will end up with CD or other GI diagnosis ⁽¹⁶⁾. Majority of the children (83%) with persistent symptoms in the LD group had LD resolution on repeat EGD. This finding is consistent with a previous conclusion by Aziz et al ⁽¹⁹⁾ that many of LD cases had no clear etiology and resolved on repeat EGD. Previous studies suggested that medication use like aspirin, NSAIDs or PPI is associated with identifying LD ⁽²⁰⁾. Similarly, our data showed that LD was more common in children with recent exposure to PPI or antibiotics, suggesting that LD may represent a transient nonspecific mucosal immune reaction to environmental, infectious, or pharmacologic triggers.

In our cohort, most of HLA testing was performed after identifying LD and it was positive in 50%, which did not seem to help and was not majorly different than the rate of celiac HLA in the general population, which has been estimated to be as high as 40% ^(21–23). Due to the limitations associated with celiac blood tests, other interventions have been tried like gluten challenge (trial of GFD with repeat SBB), which is invasive, especially with the low reported rate of CD in LD cases. Other studies suggested the use of symptom improvement on GFD to determine the potential of CD diagnosis. Symptom improvement, however, is not reliable as up to 38% of LD patients who report response to GFD are HLA-negative, and their response can be transient or because of placebo effect ⁽²⁴⁾.

On the basis of our results, we believe that IELs are a normal part of the small bowel defense mechanism and LD is more likely a transient reaction to pharmacologic, infectious, or environmental triggers ^(8,25). Many of the children with LD were diagnosed with functional GI disorders like dyspepsia, chronic abdominal pain, or irritable bowel syndrome. It is not clear if the finding of LD should change the approach or the outcome of those cases when compared with children with functional diagnoses and normal SBB histology. In our cohort, the small number of children who ended up with new GI diagnosis like IBD or CD had suggestive symptoms and abnormal biomarkers (blood and/or stool). On the basis of that perhaps, the right approach to children with LD should start with revisiting their presenting symptoms, recent medication exposure (PPI, antibiotics, and NSAIDS), and their previous investigations. In children who were not on GFD at time of EGD with benign history, no alarming signs (fever, weight loss, or bloody stools), negative investigation including blood tests (CBC, negative celiac markers and inflammatory markers), and stool studies (pathogen panel and calprotectin) in cases of diarrhea, LD should be considered a benign transient finding and reassurance with follow-up will be sufficient. On the other hand, if the child was on GFD diet or is at high risk for CD (has family history, autoimmune disorder, or chromosomal disorder) at the time finding LD, then the provider and child's family can discuss the next steps and if any further testing should be done. HLA testing results can be considered as well knowing that it will help exclude the risk of CD when negative. Conversely, in the event of positive HLA testing and/or the family is really interested in excluding CD, then 6 to 8 weeks of gluten challenge can be discussed. The child will have to consume ≥3 g of gluten for 6 to 8 weeks followed by repeating celiac markers with or without repeat EGD to r/o CD ⁽²⁶⁾. The biggest limitations to our results are the retrospective nature of the study and the availability of follow-up data.

In conclusion, the rate of LD in children is similar to reported rate in adults. In the absence of Crohn disease, CD or H pylori, LD seems to be a benign and transient histologic finding in children.