Small bowel stomas are associated with higher risk of circulating food-specific-IgG than patients with organic gastrointestinal conditions and colostomies.

OBJECTIVE: The effects of food sensitivity can easily be masked by other digestive symptoms in ostomates and are unknown. We investigated food-specific-IgG presence in ostomates relative to participants affected by other digestive diseases.
DESIGN: Food-specific-IgG was evaluated for 198 participants with a panel of 109 foods. Immunocompetency status was also tested. Jejunostomates, ileostomates and colostomates were compared with individuals with digestive tract diseases with inflammatory components (periodontitis, eosinophilic esophagitis, duodenitis, ulcerative colitis, Crohn's disease and appendicitis), as well as food malabsorption due to intolerance. A logistic regression model with covariates was used to estimate the effect of the experimental data and demographic characteristics on the likelihood of the immune response.
RESULTS: Jejunostomates and ileostomates had a significant risk of presenting circulating food-specific-IgG in contrast to colostomates (OR 12.70 (p=0.002), 6.19 (p=0.011) and 2.69 (p=0.22), respectively). Crohn's disease, eosinophilic esophagitis and food malabsorption groups also showed significantly elevated risks (OR 4.67 (p=0.048), 8.16 (p=0.016) and 18.00 (p=0.003), respectively), but not the ulcerative colitis group (OR 2.05 (p=0.36)). Individuals with profoundly or significantly reduced, and mild to moderately reduced, levels of total IgG were protected from the formation of food-specific IgG (OR 0.09 (p=<0.001) and 0.33 (p=0.005), respectively). Males were at higher risk than females.
CONCLUSION: The strength of a subject's immunocompetence plays a role in the intensity to which the humoral system responds via food-specific-IgG. An element of biogeography emerges in which the maintenance of a colonic space might influence the risk of having circulating food-specific-IgG in ostomates. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Ostomates must control their diet while managing symptomatology related to food intake.

Food-specific-IgG-based elimination diets have been shown to be effective at reducing symptoms in some digestive diseases.

Food sensitivity is a complex phenomenon that is still misunderstood.

A first analysis food-specific-IgG presents across digestive diseases or resections.

The impact of hypogammaglobulinaemia on circulating food-specific-IgG presentation.

Uncover a potential biogeography implication on food-specific-IgG risk.

These findings broaden our understanding of risk of presenting circulating food-specific-IgG in digestive diseases.

Further studies need to investigate the impact of colon resection on the immune system.

Food-specific-IgG-based elimination diet might be an option for ostomates to improve quality of life outcomes.

Introduction

From the moment that a patient undergoes digestive tract-resection surgery leading to an ostomy, they are faced with a multitude of challenges and often experience a significantly reduced quality of life.1 2 The needs for surgery are diverse and include escalation of gastrointestinal disorders such as Crohn’s disease and ulcerative colitis, cancer and associated treatment, or traumatic abdominal injury. For jejunostomates, ileostomates and colostomates, adjustments in day-to-day living must also be made in order to manage food intake linked to aspects of stoma output like volume, consistency, gas release and frequency.1 2 This activity balances nutritional intake, nutrient absorption, hydration and quality of life. Many aspects of managing output also affect the psychological well-being of ostomates, and can be amplified by personal circumstances, such as marital status or religion.1–3

While no current guidelines encourage the exclusion of specific foods from the diets of ostomates, there are recommendations for managing the volume, consistency and gas release of stomas.4 Individuals with jejunostomies and ileostomies can frequently have issues surrounding dehydration and electrolyte imbalance due to the loss of water that is normally reabsorbed in the colon.5 Deficiencies of vitamin B12, iron and zinc have also been observed in ileostomates and are associated with reduced quality of life.1 These observations extend beyond minerals and vitamins to include short chain fatty acids and polyphenols, for example.6 7 In the midst of the many issues that ostomates can face, there have been no studies on the presence of food sensitivity in ostomates.

Outside of the studies of food allergies that are, by definition, IgE mediated, the inconsistency of keywords between authors has been detrimental to a greater understanding of IgG, immune and non-immune-mediated food sensitivity and intolerance as well as food-related inflammation. These mechanisms of delayed symptomatology of food sensitivity and intolerance can also modify output characteristics. The mechanisms leading to the presence of food-specific-IgG are still under investigation in the larger context of immunotolerance, oral tolerance, inflammation, intestinal permeability, leaky gut and temporary or more long-term mucosal damage. Two challenges face us all: the difficulty to pinpointing the delayed effects of foods associated with the sensitivity process and the large amount of confusion surrounding definitions and symptomatology descriptions.

There is a growing body of evidence correlating food-specific IgG and disease or comorbidities associated with bearing an ostomy. These comorbidities include Crohn’s disease, ulcerative colitis, inflammatory bowel disease, headaches and anxiety.8–11 An IgG-guided elimination diet has shown to partially alleviate the symptoms. The designed diet can affect faecal output, digestive symptoms and quality-of-life measurements.12–15

The present work investigates food sensitivity, as defined by the presence of circulating IgG against food antigens, in participants with different digestive disorders. The rates of positivity as well as the relative intensity of response toward food antigens in ostomates were compared with samples associated with diagnostic codes of inflammatory diseases identified along the digestive tract as well as food malabsorption due to food intolerance.

Methods

Study population

Biobank samples were originally collected with the consent of Nebraska Medicine patients and consist of remaining donated blood samples from scheduled laboratory tests. The inclusion criteria for the request were for deidentified sera from individuals over the age of 19, the age of adulthood in Nebraska, with specific medical diagnoses affecting the digestive tract as described below. The exclusion criteria included the presence of a urostomy and that no two samples from the same individual were to be included. A total of 198 deidentified serum samples were acquired with the following diagnoses: appendicitis (n=18), colostomy (n=18), Crohn’s disease (n=18), duodenitis (n=25), eosinophilic esophagitis (n=15), food malabsorption due to intolerance (n=18), ileostomy (n=31), jejunostomy (n=22), periodontitis (n=18) and ulcerative colitis (n=15). Specific ICD-10 codes (International Classification of Diseases version 10) can be found in online supplemental table 1. Eosinophilic esophagitis served as positive controls based on prior knowledge.16

The public involvement, prior to this research, was done through informal discussions with the ostomate community regarding the diverse symptoms observed with different food intake. Ostomates, ostomy nurses and advocates have expressed interest in disseminating the published findings.

ELISA-based testing

Serum food-specific-IgG were evaluated using the Eagle Biosciences IgG (109 foods) ELISA Assay Kit (Catalogue number: CNS14M; Eagle Biosciences, Amherst, NH). This is a 96 well-based ELISA kit with a few related foods pooled into single wells, such as lemon and lime (online supplemental table 2). For further analysis, tested foods were placed into 16 groups according to the US Department of Agriculture Food Data Central database (online supplemental table 3). The ELISA plates were read using a BioTek Synergy H1 plate reader (BioTek, Winooski, VT). As per manufacturer protocol, a categorical score was assigned from zero to three based on the strength of response. The categorical sum was calculated as the sum of all assigned scores to each food for each individual. The number of foods positive was calculated as the sum of different food recognised by each individual.

Total IgG-based evaluation was performed using Human IgG ELISA assay (Catalogue number: EGG39-K01; Eagle Biosciences). The mean absorbance of duplicate standards and samples was calculated. For analysis, the data was used as continuous or categorised as per strength of immune competency.17–19 Individuals with total IgG level <299 mg/dL, 299–599 mg/dL, 600–1600 mg/dL and >1600 mg/dL were classified as ‘profoundly or significantly reduced’, ‘moderately reduced’, ‘normal’ and ‘elevated’, respectively.

Total IgA-based evaluation was performed using Human IgA ELISA assay (Catalogue number: HUG39-K01; Eagle Biosciences). Samples and standards were run in duplicate. For analysis, the data were used as continuous or categorised as per strength of immune competency.19 20 Individuals with total IgA level <7 mg/dL, 7–60 mg/dL, 61–356 mg/dL and >356 mg/dL were classified as ‘deficient’, ‘reduced’, ‘normal’ and ‘elevated’, respectively.

Serum calprotectin was quantified using Calprotectin ELISA kit (Catalogue number: ab267628; Abcam, Cambridge, UK). Samples and standards were run in duplicate. Serum calprotectin was used to evaluate both systemic and digestive tract inflammation at the time of sampling. For analysis, the data were used as continuous data or the calprotectin and associated levels of inflammation were categorised.21 Individuals with serum calprotectin level <215 ng/mL, 215–3800 ng/mL and >3800 ng/mL were classified as ‘low’, ‘normal’ and ‘high’, respectively.

Statistical analysis

All analyses were performed using the statistical environment R (V.4.0.3) using the Integrated Development Environment RStudio for mac OS (V.1.4.1103). All standard curves were plotted on a semilog graph, with the concentration plotted logarithmically and the optical density plotted linearly, using the R package ‘drc’.22 The best-fit line was calculated using a four-parameter logistics curve.

Statistical differences between groups (ICD-10 or categorical classification for total IgG, IgA and calprotectin) were analysed using a Kruskal-Wallis analysis of variance (ANOVA) and Dunn’s test, using the ggpubr package. Typically, when performing a Dunn’s test, a multiple testing correction is applied to the resulting p values in order to avoid an inflated type 1 error level. These adjustments are quite conservative due to the large number of groups being tested, hence p values for the Dunn’s test presented in the results section are unadjusted unless otherwise specified. Wilcoxon rank-sum test was used to assess differences in the number of foods present, categorical sum, total serum IgG, total serum IgA and total serum calprotectin.

To further investigate the possible factors (ie, covariates) impacting response, a logistic regression model was used to assess the impact of total serum IgG, total serum IgA and ICD-10 of selection on the presence of food-specific IgG while controlling for age and body mass index (BMI). The likelihood of food-specific IgG presence was used as the dependent response variable. The logistic regression model was used to estimate the degree to which ICD-10, gender, total IgG and total IgA impact the likelihood of response. Because some BMI measurements were missing from the metadata (n=46), values have been imputed for analysis using k-nearest neighbours methodology by using the kNN function of the VIM package (k=6).23 All statistical significance was determined at p<0.05.

Results

To investigate the risk of food sensitivities in ostomates, we compared 198 samples from jejunostomates, ileostomates, colostomates and other diseases localised along the digestive tract. These include, per positioning along the digestive tract, periodontitis, eosinophilic esophagitis, duodenitis, Crohn’s disease of the small intestine, appendicitis, ulcerative colitis and food malabsorption due to intolerance. The sample was composed of 52.5% females and 47.5% males, with a mean age of 49.70±17.50 years (online supplemental table 4). Eighty-three per cent of serum samples originated from Caucasian individuals, 11% from African Americans, 2% from Native Americans, 1% from Asian individuals and 3% from individuals of unspecified race.

The top 10 most prevalent food antigens detected within the samples were cow’s milk (55.56%), egg white (50.00%), wheat (36.36%), goat’s/sheep’s milk (35.35%), egg yolk (32.83%), beer yeast (28.28%), peanut (19.19%), bread yeast (18.69%), gluten (14.65%) and soybean (14.14%). A total of 55 out of 109 foods were detected, with 31 of them detected in at least 5% of the population.

The top five food categories detected were milk (55.56% of the individual tested positive for at least one product), eggs (51.52%) cereals grains and pasta (43.43%), legumes and legumes products (32.32%) and yeast (28.28%). At the exception of cereals grains and pasta, in the other four categories, all members of the category have been detected. Of note, no antigens were detected in the dark green vegetables and poultry categories. The remaining categories had only a subset of the foods detected (online supplemental table 2 and 5).

The distributions of positive foods across the different groups are similar; few foods are being shared by a significant proportion of individuals (like the most prevalent foods above mentioned) with a quick decrease in prevalence (online supplemental figure 1). Due to the over-representation of IgG against milk and egg categories in healthy individuals, they were excluded from the statistical analysis.11 24–27

The number of foods positive present in each diagnostic group was examined. There was a significant difference across the 10 ICD-10 groups (p=0.015), as shown by a non-parametric one-way Kruskal-Wallis ANOVA. To confirm this result, a Dunn’s test was performed post hoc. It indicated a significantly larger number of positive foods for those with jejunostomy vs individuals diagnosed with periodontitis (p=0.002), duodenitis (p=0.006) or appendicitis (p=0.048). Similar observations were made for ileostomates vs individuals diagnosed with periodontitis (p=0.007), or duodenitis (p=0.017), and for colostomates vs individuals diagnosed with periodontitis (p=0.023). Significance values of all pairwise Dunn’s test comparisons are presented in table 1.

Table 1

Dunn’s test pairwise comparisons between diagnostic categories of interest and the number of foods positive

	Appendicitis	Colostomy	CD	Duodenitis	EE	FM	Ileostomy	Jejunostomy	Periodontitis
Colostomy	0.182
CD	0.135	0.423
Duodenitis	0.253	0.050	0.032*
EE	0.079	0.293	0.359	0.016*
FM	0.020*	0.125	0.169	0.002*	0.291
Ileostomy	0.109	0.417	0.496	0.017*	0.341	0.139
Jejunostomy	0.048*	0.237	0.305	0.006*	0.457	0.311	0.277
Periodontitis	0.138	0.023*	0.014*	0.304	0.007*	0.001*	0.007*	0.002*
UC	0.492	0.199	0.151	0.258	0.091	0.026*	0.127	0.059	0.145

*Indicates p≤0.05.

CD, Crohn’s disease; EE, Eosinophilic esophagitis; FM, Food malabsorption; UC, Ulcerative colitis.

Similarly, the relative intensity of response to all foods was investigated using the categorical sum of response for each food per individual, in each category. A significant difference in the categorical sum per diagnostic category was observed using a Kruskal-Wallis ANOVA (p=0.013). A Dunn’s test indicated that there was a significant difference in the categorical sums of those with jejunostomy versus individuals diagnosed with periodontitis (p=0.003), or duodenitis (p=0.007); ileostomates versus individuals diagnosed with periodontitis (p=0.006), or duodenitis (p=0.014); and colostomates vs individuals diagnosed with periodontitis (p=0.029) (table 2).

Table 2

Dunn’s test pairwise comparisons between ICD-10 groups of interest and categorical sum

	Appendicitis	Colostomy	CD	Duodenitis	EE	FM	Ileostomy	Jejunostomy	Periodontitis
Colostomy	0.261
CD	0.110	0.278
Duodenitis	0.200	0.063	0.015*
EE	0.107	0.263	0.470	0.017*
FM	0.031*	0.110	0.261	0.002*	0.296
Ileostomy	0.131	0.343	0.398	0.014*	0.372	0.164
Jejunostomy	0.073	0.216	0.433	0.007*	0.467	0.307	0.321
Periodontitis	0.105	0.029*	0.007*	0.306	0.007*	0.001*	0.006*	0.003*
UC	0.409	0.201	0.081	0.290	0.079	0.022*	0.095	0.052	0.167

*Indicates p<0.05.

CD, Crohn’s disease; EE, eosinophilic esophagitis; FM, food malabsorption; UC, ulcerative colitis.

The strength of the humoral response was tested by quantifying both total serum IgG and IgA antibodies. Total serum IgG was first analysed as a continuous variable and compared with the categorical sum of food-specific IgG. A linear regression indicated a strong positive correlation (p<0.001). Classifying the same data into medically relevant groups enabled us to also show a difference between the ICD-10 groups, using a Kruskal-Wallis ANOVA (p=0.002) (figure 1). A post hoc Dunn’s test was performed, and significant pairwise differences were observed between elevated and mild-moderately reduced (p=0.03), mild-moderately reduced and normal (p=0.04), elevated and profoundly or significantly reduced (p=0.014), and normal and profoundly or significantly reduced (p=0.019). P value adjustments were made using the Benjamini-Yeukateli adjustment.

Figure 1

Significant differences were observed between the IgG overall response and the categorical sum of IgG food-specific response for each individual.

Similarly, total serum IgA was quantified and analysed first as a continuous variable. A Kruskal-Wallis test indicated that there were no significant differences in the levels of total serum IgA between ICD-10 groups tested (p=0.74). After classifying the same data into medically relevant groups, a Kruskal-Wallis ANOVA was performed and indicated that there were no significant differences between the groups (p=0.56).

To test that there was no digestive and/or systemic inflammation at the time of sampling, serum calprotectin was quantified. A Kruskal-Wallis test indicated that there were no significant differences in the levels of serum calprotectin between groups tested (p=0.72). After classifying the same data into medically relevant groups, a Kruskal-Wallis ANOVA was performed and indicated that there were no significant differences between the groups (p=0.081). As calprotectin is a measure of a transient inflammatory event, the measurement was not included in the next analysis.

To better understand the impact of demographics (age, BMI and sex), adaptive response (total serum IgG and IgA and food-specific IgG) and ICD-10 on the risk of food sensitivity, a logistic regression model was used (table 3). In this model, jejunostomates and ileostomates were, respectively, 12.70 and 6.19 times more likely to have at least one food sensitivity compared with individuals diagnosed with periodontitis. In contrast, colostomates had an OR of 2.69 and the test was shown to be not statistically significant. Interestingly, individuals with Crohn’s disease of the small intestine had an OR of 4.67, reaching statistical significance, while those with ulcerative colitis had a non-significant lower OR. Food malabsorption due to intolerance group, as well as our positive control group, eosinophilic esophagitis, showed statistical significance and large ORs. Individuals with profoundly or significantly reduced and mild-moderately reduced levels of serum total IgG were less likely relative to the normal and elevated groups to develop food-specific IgG, with ORs of 0.09 and 0.33, respectively. Total IgA levels categories did not reach significance.

Table 3

OR and 95% CI for the parameters included in the logistic regression model

Characteristic	OR	95% CI	P value
(Intercept)	0.36	0.04 to 2.82	0.330
Age	1.01	0.99 to 1.04	0.197
BMI	1.03	0.97 to 1.10	0.285
Sex
Female	1.00	—
Male	2.39	1.19 to 4.97	0.017**
ICD-10 based categories*
Periodontitis	1.00	—
Ulcerative colitis	2.05	0.45 to 10.10	0.360
Duodenitis	2.46	0.59 to 11.00	0.224
Appendicitis	2.50	0.59 to 11.40	0.222
Colostomy	2.69	0.57 to 13.60	0.216
Crohn’s disease	4.67	1.06 to 23.20	0.048**
Ileostomy	6.19	1.58 to 27.00	0.011**
Eosinophilic esophagitis	8.16	1.58 to 49.80	0.016**
Jejunostomy	12.70	2.71 to 71.60	0.002**
Food malabsorption	18.00	3.09 to 160.00	0.003**
Total IgG status
Normal	1.00	—
Profoundly or significantly reduced	0.09	0.02 to 0.33	<0.001**
Mild-moderately reduced	0.33	0.15 to 0.70	0.005**
Elevated	4.09	0.59 to 83.10	0.219
Total IgA status
Normal	1.00	—
Deficient	1.38	0.11 to 33.90	0.819
Reduced	0.33	0.06 to 1.87	0.198
Elevated	0.41	0.16 to 1.05	0.064

AIC=245.70.

*ICD-10 ordered by increasing OR value.

**Indicates p<0.05.

BMI, body mass index.

Discussion

In this cross-sectional analysis, food-specific IgG was evaluated for 198 participants for a panel of 109 foods using deidentified biobank clinical serum samples. Ostomates with jejunostomy and ileostomy showed a significant risk of presenting circulating food-specific IgG in contrast to colostomates. In addition, ileostomates and jejunostomates had significantly higher categorical sums and numbers of foods positive relative to colostomates, as compared with each other or in the context of inflammatory disease groups. To the best of our knowledge, this is the first study to link the type of digestive resection in ostomates with the risk of having circulating food-specific IgG. These findings may improve dietary management while also managing symptomatology related to food intake.

The proposed link between food-specific IgG and ostomy management is based on the effect of directed elimination diets on digestive parameters. In particular, food-specific-IgG-based elimination diets have been shown to be effective at reducing symptoms in individuals with Crohn’s disease and ulcerative colitis, as well as in people with irritable bowel syndrome.12–15 28 These designed diets affect faecal output, digestive symptoms and quality-of-life measurements.

In absence of an easily detectable issue like a food allergy, confirmed by IgE detection, identifying which food could be causing the symptoms can be a long and complex process that includes the elimination of specific foods or food ingredients, while monitoring symptoms. Faecal output appearance and consistency are important semiological descriptors for self-care and clinical management. However, until recently, there has been a lack of any output quality evaluation scale, an equivalent to a Bristol stool form scale, to monitor these signs overtime in relation to any food, medication or hydration regiment.29 30 The daily constraints associated with stoma maintenance and nutritional management may be reasons for overlooking symptoms associated with specific food intake. The recognition of specific foods that could potentially trigger symptoms could be beneficial for quality-of-life management.

There are multiple factors that may explain the observed discrepant responses between the diverse groups and ostomates. Regarding oral tolerance, antigen sampling along the digestive tract is not homogeneous or equivalent. It involves different gut-associated lymphoid tissues including Peyer’s patch and SM-ILF that are expressed at different density along the biogeography of the digestive tract, as well as variations in the tolerogenic liver environment.31–33 Individuals lacking a colon may be at a higher risk of developing food-specific IgG due to missing component of the immune system postsurgery. Alternatively, maintenance of long-term chronic digestive-tract inflammation in the ileal conduit could impact the antigen sampling process and lead to differences in response to food antigens depending on the cause of ostomy. These considerations underline that biogeography, long-term inflammation and immune tolerance sampling sites may play major roles. These hypotheses could not be further investigated due to the nature of the available samples (deidentified samples from a biobank).

The importance of general humoral immunocompetence as an indicator of food intolerance risk was supported by our results as individuals with higher levels of total IgG tended to correspond to higher levels of overall reactivity to food-specific antigens, as observed via categorical sum. In contrast, the individuals with profoundly or significantly and mild-moderately reduced levels of total IgG had significantly protective ORs of presenting circulating food-specific IgG. The total intake of a particular food may also impact the level of detectable IgG,26 however, our findings show an impact across the panel of foods included in the test.

Additional findings include the strength of the food-specific IgG response in individuals with food malabsorption due to intolerance, in categorical sum, diversity of food antigens detected and ORs. Also, individuals with Crohn’s disease of the small intestine had significantly higher numbers of foods positive and levels of overall reactivity than individuals with ulcerative colitis, duodenitis and periodontitis. Further, in the logistic regression a diagnosis of Crohn’s disease was a predictive factor for the presence of food-specific IgG.

What to think if I am an affected patient? While few foods seem to have a broad and communally shared impact on the presence of food-specific IgG (online supplemental table 5) and (online supplemental figure 1), the IgG pattern is rather scattered and personalised. Implementing an overly restrictive diet as a way to mitigate symptoms of food-related inflammation—while appealing—might likely result in undernutrition that would be significantly detrimental. The role of food-specific-IgGs testing in routine clinical practice in those undergoing ostomy surgery and in most of the disease groups mentioned is at this stage unknown yet clearly worth of investigation.

The strengths of this study reside in the choice of target groups for the analysis that comprised inflammatory diseases from the oral cavity to the colon. Eosinophilic esophagitis provided a true positive control group as the disease is well known to have an IgG component.16 We were also able to control for a range of important potential confounders, (eg, BMI, age, sex and immunocompetency), which were not included in previous analyses of food-specific IgG.

Our study also had several limitations. As the samples were deidentified, we were not able to investigate past medical history, including the main medical reason for ostomy surgery, medication intake that might influence immunocompetency, or past dietary intake. Per the design of a cross-sectional study, we do not have presurgical and postsurgical samples for the ostomates, thus, we are unable to identify if the food sensitivities were present prior to surgery. However, the needs of ostomates regarding food intake management do increase significantly postsurgery, and any tools provided to the community to improve management output are of significant importance. Lastly, for a panel of food antigens spanning 16 food categories, there is no available kit working with limited serum supplies to test IgG subtype or other classes of immunoglobulin.

In conclusion, food sensitivity risk is increased significantly in the ostomate population, and the risk is associated with the type of overall resection observed. The strength of the subject’s immunocompetence seems to play a great role in the intensity to which the humoral system responds via food-specific IgG. An element of biogeography emerges where the maintenance of a colonic space or ileal chronic inflammation influences the risk of having circulating food-specific IgG. Questions related to the effect of the immune tolerance biogeography, the strength of the adaptative immunity on food sensitivity, and the potential impact of elimination diet on the health and wellness of ostomates still need to be answered.