Micronutrient Deficiencies and Anemia in Children with Inflammatory Bowel Disease

Children with inflammatory bowel disease (IBD) are at risk of developing nutrition deficiencies, particularly because of reduced intake, restrictive diets, malabsorption, and excessive nutrient loss. The aim of this study was to determine the prevalence and predictors of anemia and micronutrient deficiencies at diagnosis and one year follow up in children and adolescents with inflammatory bowel disease (IBD). Children and young adults diagnosed with IBD before the age of 17 years between 2012 and 2018 were included. Laboratory measurements including serum levels of iron, ferritin, zinc, vitamin D, vitamin A, vitamin E, selenium, copper, vitamin B12, and red blood cell (RBC) folate at diagnosis and one-year follow-up were documented as part of the Manitoba Longitudinal Pediatric Inflammatory Bowel Disease (MALPID) Cohort. A total of 165 patients with IBD were included, 87 (53%) with Crohn’s disease (CD) and 78 (47%) with ulcerative colitis (UC). The prevalence of deficiencies in our cohort at diagnosis and one year follow-up, respectively, were iron (56% and 27%), ferritin (39% and 27%), zinc (10% and 6%), vitamin D (22% and 13%), vitamin A (25% and 25%), vitamin E (5% and 4%), selenium (10 and 7%), copper (17% and 27%), vitamin B12 (2% and 5%), and Red blood cell (RBC) folate (1% and 17%). Anemia was present in 57% and 25% at diagnosis and follow up respectively. In CD patients, age of diagnosis (15y–younger than 18y) was a predictor of moderate to severe anemia and albumin levels (<33 g/L) were protective against anemia. Many children with IBD suffer from anemia and micronutrient deficiencies at diagnosis and some fail to recover after one year despite being in clinical remission.


Introduction
Crohn's disease (CD) and ulcerative colitis (UC), collectively known as inflammatory bowel disease (IBD), are non-curable chronic inflammatory disorders of the gastrointestinal tract. Approximately 10-25% of cases begin during childhood or adolescence, and their incidence appears to be increasing [1].
Pediatric patients with active IBD are at higher risk for micronutrient deficiencies through several different mechanisms, including suboptimal oral intake, nutrient malabsorption, increased intestinal losses, systemic inflammation, hypermetabolic state, and medication's adverse events [2][3][4]. Intestinal inflammation in pediatric IBD is associated with malabsorption, maldigestion, and gastrointestinal protein loss, contributing to deficiencies of energy, protein, and micronutrients [5]. Inflammatory mediators specifically interfere with the absorption or utilization of certain nutrients, especially iron and vitamin D [6,7].
Nutritional status has been shown to be an essential factor in determining the prognosis of IBD [8]. Micronutrient deficiencies have been shown to have significant implications on the outcomes of patients with IBD especially in those with anemia with subsequent lower quality of life and cognitive dysfunction [9]. A recent systematic review, which included 39 pediatric studies, concluded that iron and vitamin D deficiencies are common in pediatric patients with IBD, whereas vitamin B12 and folate deficiencies are rare [10]. In another study, zinc deficiency occurred at a higher rate in patients with CD than in healthy controls [11].
Anemia is a common problem in pediatric IBD patients and is reported in up to 75% of patients [12,13]. Iron deficiency anemia (IDA) is the most common form of anemia due to the lack of sufficient iron to form normal red blood cells. Iron deficiency anemia is typically caused by inadequate intake of iron, chronic blood loss, or a combination of both [14]. Serum iron alone is an unreliable marker of iron deficiency as it is influenced by a variety of factors including diurnal variation, inflammatory processes (decreased), malignancy (decreased) and menstrual blood loss (decreased). For this study, we have defined anemia using the World Health Organization (WHO) guidelines, as a decline in blood hemoglobin based on age and sex in children. Few studies that examined anemia have addressed potential predictors or factors associated with anemia such as age, sex, family history of IBD, clinical disease activity, and inflammatory biomarkers, such as high C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and albumin.
We aimed to identify serum micronutrient status at diagnosis and one-year follow-up, anemia prevalence at diagnosis and after a one-year follow-up, and factors associated with anemia at diagnosis and follow-up.

Materials and Methods
Our longitudinal, population-based cohort comprised all children and young adults (<17 years) diagnosed with IBD in the Canadian province of Manitoba between January 2011 and August 2018 and who consented to be enrolled in the Manitoba Longitudinal Pediatric Inflammatory Bowel Disease (MALPID) Registry [15][16][17]. Patients were recruited from Winnipeg Children's Hospital, which is the only pediatric tertiary-care center in the province of Manitoba. Data on anthropometry, prescribed medications, inflammatory markers, and disease characteristics were jointly retrieved from the hospital electronic patient database and reviewed their medical and dietetic notes. All patients were diagnosed according to established clinical, endoscopic, histological, and radiological guidelines [18]. Disease location and behavior, growth impairment and perianal involvement for Crohn's disease, and disease extent for ulcerative colitis are presented according to Paris classification [19].

Description of Variables
Demographic, anthropometric, clinical, laboratory, radiological, and endoscopic data were recorded for each patient at diagnosis. Height, weight, and body mass index were converted to age and sex-adjusted standard deviation scores (Z scores) using the Centre for Disease Control growth reference charts [20]. Clinical disease activity was assessed using the Physician Global Assessment (PGA) scores at diagnosis and follow up. Disease phenotype at diagnosis was categorized according to the Paris Classification [19]. Data on nutrition variables were retrieved at diagnosis (±) 7 Days, and after one year (±) 3 months. Each parameter was described both as a categorical (within normal limits or deficient) and nominal value. Anemia was defined according to age and sex normal values using WHO guidelines [21]. As per the WHO guides (hemoglobin g/L), for children 6-59 months of age; mild anemia is 100-109, moderate is 70-99 and severe is lower than 70. For children 5-11 years of age, mild anemia is 110-114, moderate is 80-109, and sever is lower than 80. For children 12-14 years of age, mild anemia is 110-119, moderate is 80-109 and severe is lower than 80. For 15 years of age and above females (non-pregnant) non-anemia is 120 or higher, mild is 110-119, moderate is 80-109 and severe is lower than 80. For 15 years of age and above males, non-anemia is 130 or higher, mild is 110-129, moderate is 80-109, severe is lower than 80. Ferritin levels were adjusted using the ECCO guidelines that recommend adjusting serum ferritin concentration by concurrently measuring C-reactive protein (CRP) to remove effects of subclinical inflammation [22].

Statistical Methods
The sample size calculation for this study was based on the assumption that the majority of patients would participate in the study and the proportion of patients with any micronutrient deficiency between time points or between type of disease would differ by 20%. The sample size was calculated based on a log-rank test and required at least 121 patients at each time point for each disease (CD or UC), using significance level of 5% and power of 80%. Multiple imputation by chained equations (MICE) method was used to fill in 15% of partially missing data. Twelve patients were excluded from the analysis because of incomplete initial data.
Potential discriminants of anemia incidence and the extent of its severity were defined a priori at disease diagnosis and at one-year follow-up. At diagnosis and one year followup these included disease type, disease phenotype in CD, extensive colitis in UC, age at diagnosis, gender, systemic biomarkers of disease activity, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and serum albumin (AB), and family history of IBD (FH of IBD).
Continuous variables were presented either with means and SD or with medians and interquartile range (IQR) depending on the data's distribution. Normally distributed continuous variables were described as mean and standard deviations, whereas non-normally distributed continuous variables were reported as medians and interquartile ranges (IQRs). Continuous variables were compared using simple independent t-tests or Mann-Whitney tests, whereas categorical variables were compared using Chi squaretests or Fisher-exact tests. Correlations between continuous variables were evaluated using Spearman p correlation or Pearson coefficients, as appropriate. Differences between groups were assessed with a 2-sample t-test and analysis of variance for parametric variables. Correlations between continuous variables were measured using Spearman's rho correlation. All statistical tests were two-tailed, and p < 0.05 was considered statistically significant. Multivariate ordinal logistic regression was used to analyze predictors of anemia at diagnosis and one-year follow-up. For the purposes of analysis, moderate, and severe anemia types were combined, physician global assessment (PGA) moderate and severe were combined and referred to as active disease. Age of diagnosis categories included 0-younger than 12y, 12-younger than 15y, and 15-younger than 18y. For the predictive analysis at diagnosis CD and UC cohorts were modelled separately, at follow up they were modelled together. Hosmer-Lemshow test was used to calculate goodness of fit, Mcfadden pseudo R-squared was used to calculate predictive strength and the positive and negative predictive values (PPV and NPV) were calculated for anemia non-recovery at follow-up. StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LPStatistical (STATA, College Station, TX, USA) was used for the analysis.

Ethics
The study's protocol (HS20359 H2016:456) was approved by the Health Research Ethics Board at the University of Manitoba, Winnipeg, MB, Canada.

Crohn's Disease
A total of 87 children (40 girls) with CD were included. The median age at diagnosis 13 (IQR: 10-15) years.
Serum levels of specific vitamins and minerals at diagnosis and after one year of treatment are summarized in Table 2.
We found a significant positive correlation between serum iron deficiency and copper deficiency and at diagnosis (p = 0.038) and with serum iron deficiency and zinc deficiency (p < 0.001). There was also a statistically significant positive correlation found between B12 deficiency and RBC folate deficiency (p = 0.0267). There were no significant associations between duodenal involvement (L4a) and anemia or RBD folate deficiency, or Ileal (L1) and B12 deficiency. 1 Paired t-test was run to analyze for significant differences between micronutrient levels at diagnosis and follow-up; * Correlation is significant at the 0.05 level (2-tailed). 2 (Table 3) A total of 43 (49.0%) patients were prescribed iron supplementation at diagnosis. Of those individuals, 27 (31.0%) received oral iron compounds, 14 (16.1%) with iron salts, and two (2.3%) intravenous iron infusion. Further, 41 (47%) patients were prescribed other supplement types, which included vitamin D and a multivitamin, and 24 (27.3%) were prescribed vitamin D supplements, 11 (12.5%) were prescribed vitamin D and multivitamin (included folic acid), and six (6.8%) were prescribed a multivitamin. Supplement adherence was not captured. At diagnosis, eight (9.2%) of patients were treated with corticosteroids, 47 (54%) with exclusive enteral nutrition, 24 (28%) azathioprine (AZA), and eight (9%) with upfront anti-tumor necrosis factor (TNF) therapy. At follow-up, 38 (44%) patients were treated with azathioprine, 25 (29%) with anti-TNF medications, nine (10%) with methotrexate, and nine (10%) with combined immunomodulatory and biologic treatment.

Inflammatory Biomarkers
At diagnosis, there was a positive association between anemia and PGA (active disease) (p = 0.003), and high ESR (p < 0.001). At one-year follow-up, there were no significant associations found.

Anemia at Follow-Up
Briefly, 21 (24%) patients were anemic at one year follow-up. Further, 17 patients (20%) who were anemic (mild, moderate and severe) at IBD diagnosis (n = 87) remained so one year after diagnosis, whereas five of the 33 children who were not anemic at diagnosis had anemia at one-year-follow-up. Of those who did not recover at follow up, 10 (11%) patients had mild anemia, and seven (8%) moderate to severe anemia, with a median age of 3 (IQR: 2-4) years. Moreover, 15 (17%) patients were in remission, and two (2%) had active disease. There were no significant predictors for patients who did not recover from anemia at follow-up of ( Table 7).

Ulcerative Colitis
A total of 78 children (40 females) with UC were included. The median age at diagnosis 14 years (IQR: [11][12][13][14][15]. Serum levels of specific trace elements, minerals, and vitamins at diagnosis and after one year of treatment are presented in Table 1.
There was a significant positive correlation between copper deficiency and serum iron deficiency at follow up (p = 0.017). The status of serum micronutrients was not significantly associated with clinical disease activity at diagnosis or follow-up, with the exception of UC patients with zinc deficiency and active disease at diagnosis (p = 0.039). There was no association between 5-ASA treatment and RBC Folate deficiency.

Anemia
The mean hemoglobin concentration at disease diagnosis was 114.7 ± 21.9 g/dL and 126 ± 15.5 g/dL at follow-up. The prevalence of anemia decreased from diagnosis to one-year follow-up from 52% to 27.0% (p < 0.001). At diagnosis, 15 (19.2%) were mildly anemia, and 26 (33%) suffered from moderate to severe anemia. Anemia type by age of diagnosis, and sex is summarized in Table 5.
Family History of IBD: Of the 36 (46%) patients who did have a family history, 14 (39%) were first degree, 13 (36%) were second degree, and 10 (29%) were others. Family history of IBD was not a significant predictor of anemia at diagnosis or one-year follow-up.

Inflammatory Biomarkers
Positive correlations were found with anemia and active disease (PGA), CRP (>5 mg/L) and ESR (>10 mm/h) at diagnosis and follow-up. There was a significant association between anemia and ESR (>10 mm/h) at diagnosis (p = 0.012) and but not at follow-up (p = 0.38).  (Table 6).

Anemia at Follow-Up
Briefly, 21 (27%) patients were anemic at follow-up. Further, 15 (19%) patients who were anemic (mild, moderate and severe) at IBD diagnosis (n = 78) remained so one year after diagnosis. Five of the 38 children who were not anemic at diagnosis were anemic at follow-up. Of those who did not recover at follow up, seven (9%) patients had mild anemia, and eight (10%) moderate to severe anemia. Moreover, 13 (17%) patients were in remission, and two (3%) had active disease. There were no significant predictors of anemia non-recovery (Table 7).

Discussion
We found that both CD and UC patients had multiple nutritional deficiencies at diagnosis with the majority of have improved at follow-up. Over half of CD and UC patients had anemia at diagnosis with moderate improvement at one-year follow up. At diagnosis, there was a significant association between PGA and anemia, but not at follow-up. The status of serum micronutrients was not significantly associated with clinical disease activity at diagnosis or follow-up, with the exception of serum iron levels and vitamin A at diagnosis for the entire cohort (CD and UC), and for UC patients with zinc deficiency and active disease at diagnosis. However, some deficiencies in micronutrients were present even in states of clinical remission or quiescent disease at follow-up. Time to diagnosis is potentially an important factor for the development of anemia and micronutrient deficiencies. Patients who experience a longer period of symptoms prior to diagnosis could potentially be at a higher risk of deficiencies [23]. Disease activity in the proximal jejunum (as part of L4 phenotype) may also contribute to iron deficiency directly through decreased iron absorption. There were no significant correlations found between L4 and anemia in our study.
The high prevalence of serum iron deficiency correlated with rates of iron deficient anemia found in our study with both CD and UC, despite 73 (44%) CD and UC patients having received an iron supplement at diagnosis. In some studies, including ours, iron deficiency anemia is common in patients with IBD, even under complete remission, and iron sufficiency was not sustained following iron treatment due to ongoing inflammatory activity [11,24]. ECCO guidelines state the use of intravenous iron is the recommended route for moderate and severe disease, while oral iron has limited benefits [22]. In our study, the type of iron supplementation was not always included in patient notes. This may have led to an increase of reported oral iron supplementation.
Anemia is prevalent in this cohort of children with IBD, with 64% at presentation, falling to 25% at one year. At diagnosis, it is proportionally higher in those with CD versus UC (61% versus 52%). The prevalence or severity of anemia was similar among the types of IBD at diagnosis (moderate to severe anemia: CD versus UC, 40% versus 33%); mild anemia: CD vs. UC, 21% vs. 19%. Proportionally, more girls than boys were moderate to severely anemic (boys vs. girls: 15% vs. 21%; and more boys than girls had mild anemia: boys vs. girls, 23% vs. 7%). Age of diagnosis (15 < 18) and albumin levels (<33 g/L) were predictors of moderate and severe anemia at diagnosis, and only with the CD cohort. One reason could be that adolescents might be more private about their IBD symptoms, which could delay diagnosis and increase risk for anemia.
In our study, vitamin D levels followed previous studies of patients with CD and UC who had similar rates or underwent minimal improvement from diagnosis to followup [25][26][27]. This suggests that vitamin D status should be addressed with a more aggressive therapeutic approach. One study found that vitamin D deficiency was more common in newly diagnosed patients [25], which was also what our study found. Recent studies have investigated the relationship between vitamin D deficiency and disease severity in children with IBD, while some limited data suggested an association of vitamin D deficiency with a more severe course of disease, other studies did not report such a relationship [27][28][29]. Our study showed no statistically significant associations between vitamin D deficiency and active disease at diagnosis or follow-up. Dosage of vitamin D was not consistently captured in our cohort and thus was omitted. Additionally, patients' low compliance with vitamin D supplementation intake may be attributed to vitamin D deficiency rates during follow-up. Two pediatric IBD randomized controlled studies did not find significant differences in the effect of different oral doses of Vitamin D ranging from 400 IU to 2000 IU [26,30], whereas one RCT did show benefit for higher range doses (either 2000 IU daily or 50,000 IU) weekly [31]. The Winnipeg IBD clinic generally prescribes 1000 IU daily. However, this information was not consistently captured in the patient charts and not included in the data.
Three studies have found low serum vitamin A levels in children with IBD, consistent with our findings. One study found that 16% of patients had low serum levels [32] and another study found that 14% of children with IBD had low serum vitamin A [33]. Another study found that 40% of patients with active CD had low serum vitamin A compared with <5% of patients with inactive CD [34]. For vitamin E, our results are consistent with two existing studies. One study reported that 6% of patients with IBD had low serum vitamin E [33], while another reported 70% with active disease, 15% with inactive disease, and 5% of controls with low vitamin E [34]. Active disease was not associated with vitamin A or vitamin E deficiencies.
Low vitamin B12 levels were rare in our study at diagnosis and at follow up. However, vitamin B12 serum levels are not always an accurate assessment of B12 as serum B12 can be false normal or false high values, even if a deficiency is present [35]. One study found 22.2% of patients with CD and 7.5% of patients with UC had low vitamin B12 status and found no difference in serum B12 concentration [36]. They also reported that patients with ileal or ileocecal resection were more likely to have abnormal serum B12 concentration; this information was not included in our study. Disease location can play a role in B12 absorption due to increased inflammation in the Ileal (L1) phenotype [37]. Energy intake is linked to disease localization in CD patients, with a reduction of energy intake only in ileal (L1) and ileocolonic (L2) disease [38]. We examined CD patients with Ileal disease and B12 deficiencies at diagnosis and follow-up and found no significant differences.
In our cohort there was a significant increase in RBC folate deficiency from diagnosis to follow-up. For CD patients, it was particular high increasing from 0-24% from diagnosis to one year follow-up. Several reasons may contribute to folate deficiency. Upon diagnosis, some IBD patients may switch to a gluten-free diet, which could increase risk for RBC Folate and B12 deficiency [39]. Some medication interactions, including 5-ASA and methotrexate, reduce cellular uptake of folate [40]. In our UC cohort, there was no association between 5-ASA treatment and RBC Folate deficiency. However, there was a statistically significant positive correlation found between B12 deficiency and RBC folate deficiency in our UC cohort, but not in the CD cohort. For the CD cohort, there was no significant association between L4 phenotype, RBC folate deficiency or anemia.
The prevalence of zinc deficiency found in our study at diagnosis was 14% in CD and 6% in UC, which is lower than reported in previous studies. Other smaller cohorts found that zinc was deficient in up to 20% of patients, mainly in those with CD [7,32,34]. Several factors can contribute to zinc deficiency in IBD, including low oral intake or poor absorption well as inflammation can be a catalyst for elevated urinary excretion of zinc. In our study there were no significant correlations between zinc deficiency and CRP, ESR, albumin, or clinical disease activity.
The prevalence of selenium deficiency in our cohort at diagnosis and follow-up was 10% and 7%; and copper was 17% and 27%. For the CD cohort, we found a significant positive correlation between copper deficiency and serum iron deficiency at diagnosis and with zinc deficiency. For the UC cohort, there was a significant positive correlation between copper deficiency and serum iron deficiency at follow up. Copper is essential for absorbing iron from the gut [41], and when copper levels are low, the body may absorb less iron. Existing studies did not consistently find low selenium serum levels in the populations. One of these studies included a small sample size (n = 24) with normal mean selenium within normal limits for patients with IBD and controls [42]. Two small pediatric cohort studies yielded contradictory results [7,43].
The strengths of our study include the prospective cohort analysis that included a relatively large number of children and adolescents at diagnosis and follow-up. The number of micronutrients examined in this study significantly add to the current limited literature There are several limitations to our study including lack of a control group and data regarding the patients' dietary intake, time to diagnosis, and adherence to prescribed supplements (iron, vitamin D, multivitamins) and the type of supplementations. The frequency of using intravenous iron might have been under-reported in our cohort. This paper is focusing on serum levels of micronutrients and we could not find obvious clinical symptoms related to zinc, selenium or copper deficiencies. However, some symptoms like fatigue could be multifactorial and it is impossible to be certain if fatigue could be related to a specific micronutrient deficiency. Fatigue is common symptom in patients with IBD even in those in clinical remission and without anemia. Several causes of fatigue in IBD have been examined in the literature including inflammatory cytokines, nutritional deficiency, altered metabolism and psychological comorbidity. Different subtypes of fatigue could be related to different mechanisms including a possible role for bidirectional communication between the gut and central nervous system (the gut-brain axis) [44].

Conclusions
This study suggests that vitamin and mineral deficiencies and anemia are significant on-going issues in the care of children with IBD and highlights the scale of the problem even in children under follow up. Reliable diagnostic criteria are needed to help identify children who may be suitable for therapeutic intervention that, in turn, may make significant improvements in quality of life.
In general, there are limited studies in children with IBD and micronutrients, resulting in a lack of pediatric-generated medical evidence. The data could be used for screening purposes or therapeutic management. Given that early and sustained improvements in nutritional health may provide life-long advantages to children with IBD, in addition to potentially decreasing long-term healthcare costs, the medical and funding communities should recognize the importance of well-developed nutritional studies in pediatric IBD.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study Data Availability Statement: The data underlying this article were provided by permission from patients and approved by the Health Research Ethics Board. Due to privacy agreements line item data cannot be shared.

Conflicts of Interest:
The authors declare no conflict of interest.