Patients with inflammatory bowel disease (IBD) are potentially at risk for developing fat-soluble vitamin deficiency because of decreased nutrient intake, steatorrhea, or increased gastrointestinal losses during disease flares. In 1984, Main et al. demonstrated low plasma retinol levels in 10 of 52 patients with Crohn's disease (CD), 2 of whom had demonstrable night blindness (1). Subsequent studies have identified low serum vitamin A and/or vitamin E levels in a subset of adults with CD or ulcerative colitis (UC) (2-4). Monitoring of fat-soluble vitamin levels during nutritional restitution of the child with IBD has been recommended if “significant steatorrhea is present”(5). However, no studies have been performed to determine the frequency of fat-soluble vitamin deficiency in children with IBD and whether routine monitoring of fat-soluble vitamin levels is warranted. We therefore undertook a prospective study to determine the prevalence of low vitamin A and E serum levels in our IBD population and to identify patients at risk for the development of hypovitaminosis.
PATIENTS AND METHODS
Ninety-seven patients with inflammatory bowel disease (61 with CD, 36 with UC) under observation in the inflammatory bowel disease program at Children's Hospital, Boston, were prospectively studied during outpatient follow-up or during inpatient treatment. The diagnosis of CD or UC was based on standard clinical, radiographic, and endoscopic criteria. Data gathered prospectively at the time of evaluation included state of well-being, presence or absence of abdominal pain, stool frequency, presence or absence of hematochezia, physical examination, complete blood count, sedimentation rate, and serum albumin level. Four children with IBD (3 UC, 1 CD) had evidence of intrinsic liver disease (3 with sclerosing cholangitis, 1 with cryptogenic cirrhosis); none were jaundiced or had pruritus.
Twenty-three children and young adults with gastrointestinal illness but no evidence of inflammatory bowel disease (12 with recurrent abdominal pain, 9 with gastroesophageal reflux disease, 1 with irritable bowel syndrome, and 1 with idiopathic oral ulceration) served as a control population.
Weight and height were obtained at the time of each patient's assessment. For patients less than 18 years old, the weight-for-age z score was determined using EPI-INFO anthropometric software (6). Body mass index (BMI; in kilograms per square meter) was determined for all patients. Although z scores for body mass index in children are not available, Must et al. have published reference data defining the 15th percentile of BMI in healthy children (7). We therefore defined as underweight those individuals whose BMIs were below the 15th percentile for their age and sex.
Disease severity for patients with CD at a given visit was quantitated using the Pediatric Crohn's Disease Activity Index (PCDAI) of Hyams et al.(8) A PCDAI score of 0 to 10 denotes inactive disease; 11 to 30, mild disease; and 31 or greater, moderate to severe disease. Disease severity for UC patients was assessed using the clinical score of Kozarek et al. (9) This scoring system derives a clinical score ranging from 0 to 15, based on five parameters: number of liquid stools daily, degree of rectal bleeding, number of extraintestinal symptoms, abdominal pain, and overall well-being. The clinical characteristics of our IBD population and control group are summarized in Table 1, and medications taken by IBD patients at the time of study are given in Table 2. A subset of patients were taking a daily over-the-counter multivitamin, which contains 5000 IU of vitamin A and 30 IU of vitamin E.
One venous blood sample was obtained from each patient. Serum obtained was stored frozen at -70 °C and protected from light until the vitamin assays were performed. Vitamin A (retinol) and vitamin E (alpha-tocopherol) were simultaneously measured by a reversed-phase, high-performance liquid chromatographic method at wavelengths of 325 and 288 nm, respectively, after extraction with hexane, using an LC-10A liquid chromatograph, an SCL-10A liquid controller, an SPD-10A UV/V is detector, an SIL-10A autoinjector, and a CR-501 Chromatopac integrator/recorder (Shimadzu Scientific Instruments, Columbia, MD, U.S.A.) and a Biophase ODS column (Bioanalytical Systems, West Lafayette, IN, U.S.A.) (10). Assays were standardized, using calibrators from the National Institute of Standards and Technology. The assays demonstrated day-to-day variability (expressed as the coefficient of variation) of less than 5% for vitamin A, at concentrations of 20 and 53μg/dl, and of less than 4% for vitamin E, at concentrations of 3 and 11.3 mg/l. On the basis of established reference intervals for serum retinol and alpha-tocopherol in children, we defined a low serum retinol level as less than 20 μg/dl, and a low serum alpha-tocopherol level as less than 5 mg/l(11).
Results are expressed as mean values ± the standard error of the mean (SEM). Continuous data were compared with an independent samples Student's t-test. Discrete or proportional data for 2 × 2 contingency tables were analyzed with Fisher's exact test or the chi-square test for independence with Yates's correction (12). The Pearson correlation coefficient (r) was used to measure the linear association between continuous variables.
Statistical modeling with logistic regression was used to identify univariate and multivariate predictors of low serum vitamin levels, with the outcome variable being an individual with either hypovitaminosis A (vitamin A level <20 μg/dl) or hypovitaminosis E (serum level <5 mg/l). For the univariate analysis, regression parameters were determined using the maximum likelihood estimation, with the likelihood ratio test used to assess significance of each variable. The risk of each outcome was approximated by the odds ratio, with 95% confidence intervals around the odds ratio. Two-sidedp values < 0.05 were used as the criteria for statistical significance. For the multivariate analysis, candidate variables were entered into the logistic regression equation, with an entry criterion of p< 0.1, and a criterion for staying in the final model of p < 0.05. Fit of the final multivariate model was tested with the Hosmer-Lemeshow goodness-of-fit statistic (13). Data analysis was performed with the SPSS version 6.1 (SPSS, Chicago, IL, U.S.A.) software package.
Informed consent for phlebotomy was obtained from each participant under a protocol approved by the Children's Hospital Committee on Clinical Investigation.
Demographic characteristics for the CD patients, UC patients, and control subjects are given in Table 1. The CD group had a lower weight-for-age z score (-0.39) than the UC and control populations, but this difference was not statistically significant. In CD patients, UC patients, and control subjects the mean serum retinol levels ±SEM were 36.3 ± 2.3 μg/dl, 38.1 ± 2.8 μg/dl, and 35.8 ± 2.5μg/dl, respectively; mean alpha-tocopherol levels were 8 ± 0.42 mg/l, 7.9 ± 0.39 mg/l, and 7.5 ± 0.37 mg/l, respectively.
Prevalence of Hypovitaminosis A or E
No control children had low serum levels of vitamins A or E. In contrast, hypovitaminosis A (serum level <20 μg/dl) was present in 14 (14.4%) of the 97 IBD patients, and hypovitaminosis E (serum level <5 mg/l) was present in 6 (6.2%) (Figs. 1 and 2). Sixteen (16.4%) IBD patients had low serum levels of vitamin A or of vitamin E.Figure 3 depicts a significant linear correlation between serum vitamin A levels and vitamin E levels in the IBD patients (r = 0.35; p < 0.001). Of the 6 children with low vitamin E levels, 4 had low levels of both vitamins A and E; all 4 had CD involving the small bowel. The prevalence of hypovitaminosis A or E was similar in the CD and UC populations (Table 3).
In the inflammatory bowel disease population, 22 patients were taking a multivitamin supplement; 73 were not taking supplements. We found no difference in the prevalence of low levels of vitamins A or E between these two groups. We also found no difference in the prevalence of hypovitaminosis in underweight (BMI lower than the 15th percentile) IBD patients compared with that in patients with BMI higher than the 15th percentile.
Disease Activity and Risk of Hypovitaminosis in Crohn's Disease Patients
We examined which clinical and laboratory factors were associated with low serum levels of vitamins A or E in our CD population. We defined our primary outcome variable as patients with low serum vitamin A or low serum vitamin E levels, so that we could broadly identify any patients at risk for hypovitaminosis. We first performed a univariate analysis between candidate risk factors and probability of hypovitaminosis A or E. Significant predictors(p < 0.05) of the presence of hypovitaminosis were the PCDAI score, the erythrocyte sedimentation rate (ESR), and the serum albumin level; nonsignificant covariates included age, sex, BMI, disease location, hematocrit, and multivitamin supplementation.
There was a strong relationship with disease severity and frequency of hypovitaminosis (Table 4). In the CD population, only 1 of 26 patients with inactive disease (as defined by a Hyams PCDAI score of≤10) had a low serum levels of vitamins A or E. However, 19% of CD patients with mild activity and 36% of patients with moderate to severe activity had a low vitamin A or vitamin E level (p < 0.05 by chi-square test).
The ESR and the serum albumin level were also statistically significant predictors of the presence of hypovitaminosis A or E. There was an inverse linear relationship between ESR and serum vitamin A (Fig. 4), and higher levels of ESR correlated with an increased likelihood of low levels of vitamins A or E; specifically, 9 of 35 patients with an ESR of more than 25 mm/hour versus 1 of 26 patients with an ESR less than 25 mm/hour had low serum levels of vitamins A or E. Hypoalbuminemia also increased the likelihood of hypovitaminosis; 4 of 10 patients with albumin concentrations of less than 3 gm/dl had low serum vitamin A or vitamin E, compared with 6 of 51 patients with albumin levels of more than 3 gm/dl.
Many of the significant predictive factors of hypovitaminosis identified in our univariate analysis are colinear (active CD correlates with low hematocrit, low serum albumin level, and a high ESR). To identify which covariates were independently predictive of hypovitaminosis, a multivariate logistic regression analysis was performed. All candidate predictors were introduced into the logistic regression equation. In the multivariate analysis(Table 5), PCDAI and serum albumin were the only statistically significant independent predictors of the presence of hypovitaminosis A or E after adjusting for other covariates.
Risk Factors for Hypovitaminosis in Ulcerative Colitis
The prevalence of low serum vitamin A or vitamin E levels in patients with UC was 42.8% (3 of 7 patients) in moderate to severe colitis, 0% (0 of 10 patients) in mildly active colitis, and 15.8% (3 of 19) in patients with inactive colitis. Of note, 2 of the 3 patients with inactive UC and hypovitaminosis had intrinsic liver disease (1 patient with primary sclerosing cholangitis, 1 with cryptogenic cirrhosis) which may have predisposed them to fat malabsorption and hypovitaminosis. The prevalence of hypovitaminosis A or E was significantly different between the group with moderate to severe colitis and the groups with mild and inactive disease. We identified no other clinical or laboratory parameters that predicted low serum levels of vitamins A or E in UC. However, the statistical power is low; consequently, inferences are difficult, given the smaller sample size of the UC population.
Our study identified a 16% prevalence of low serum vitamin A or vitamin E levels in our pediatric inflammatory bowel disease population, with hypovitaminosis A more than twice as common as hypovitaminosis E. There was a direct linear correlation between vitamin A and vitamin E levels(Fig. 3), and only 2 of 97 IBD patients who had a normal vitamin A level had a low vitamin E level. Although no patients had overt complaints suggestive of deficiency, the high prevalence of hypovitaminosis suggested that children with CD or UC may be a population at risk for the development of vitamin deficiency.
On further investigation, we found that hypovitaminosis A or E is rare(<5% prevalence) in children with inactive IBD and no intrinsic liver disease, suggesting that routine measurement of levels of vitamins A or E is not necessary in healthy pediatric outpatients with inactive CD or UC. In contrast, we found a striking relationship between disease activity and low levels of vitamins A or E. Laboratory markers suggestive of disease activity(high sedimentation rate and hypoalbuminemia) also correlated strongly with an increased likelihood of hypovitaminosis. Our results are comparable to those of Schoelmerich et al., who demonstrated in adults that vitamin A levels inversely correlated with disease activity (14).
The prevalence of low serum retinol levels in adults with CD has been reported to range as high as 20% (1-3). In our pediatric IBD population, we found a similar prevalence of hypovitaminosis A as that found in adults with CD. Other studies have established that serum retinol levels of less than 20 μg/dl may place a patient at risk for physiologic sequelae of vitamin A deficiency (including impaired lung regeneration, night blindness, and corneal and conjunctival abnormalities)(1, 15-17). Further study with more sophisticated techniques (including formal dark-adaptation screening and relative dose-response assays) will be necessary to identify whether the hypovitaminosis noted in our patient population represents clinically significant deficiency (18).
Potential etiologies of low serum fat-soluble vitamin levels in children with IBD include increased consumption of antioxidants by proinflammatory free radicals, decreased nutrient intake, malabsorption, enteric losses, or altered mobilization during inflammation(19-23). Lowered serum retinol levels have been reported in children with acute diarrhea, pulmonary exacerbations of cystic fibrosis, measles virus infection, and varicella infections (18,24-26). Proposed mechanisms for depression of serum vitamin A in these acute inflammatory states include fecal losses, impaired synthesis of retinolbinding protein or increased urinary excretion of circulating retinol.
Limitations of our study were that we did not perform dietary evaluations and food frequency questionnaires in our population. Although most patients surveyed were on regular diets, there may have been some nutrient restrictions in our CD patients, given their lower mean weight-for-age z scores. In addition, we did not measure retinolbinding proteins or prealbumin, because these proteins are not commonly obtained in clinical practice to screen for vitamin A deficiency, and the retinol binding level generally correlates with serum levels of vitamin A. Finally, we did not measure serum lipids and calculate vitamin E/lipid ratios. Although vitamin E/lipid ratios may be more accurate measures of vitamin E status, serum alpha-tocopherol is the more commonly used screening test. Further studies are necessary to ascertain whether these additional tests will be useful in the clinical management of IBD patients.
In conclusion, low levels of serum retinol and alphatocopherol are more common than previously suspected in children with CD and UC. There is a positive correlation between serum levels of vitamin A and vitamin E in IBD, and hypovitaminosis E is rare in the absence of hypovitaminosis A. Children with inactive disease rarely have hypovitaminosis, but children with active CD or UC have a 30% to 40% prevalence of low A or E levels. In the future, we plan to determine whether these low serum levels reflect true vitamin deficiency or are epiphenomena of inflammation and whether vitamin supplementation corrects hypovitaminosis in individuals with active IBD.
Acknowledgment: The authors thank the physicians and nurses of the gastroenterology division for referring patients for the study, and Donna Lee for assisting with the vitamin assays.
Supported in part by a National Institutes of Health (Bethesda MD, U.S.A.) General Clinical Research Centers Clinical Associate Physician Award (2MO1 RR 02172 [AB]) and by a National Institutes of Health grant ([CD] P30-DK40561; Clinical Nutrition Research Center at Harvard).
1. Main AN, Mills PR, Bronte-Stewart J, Nelson LH, McLelland A, Shenkin A. Vitamin A
deficiency in Crohn's disease
1983; 24: 1169-75.
2. Fernandez-Banares F, Abud-Lacruz A, Xiol X, et al. Vitamin status in patients with inflammatory bowel disease. Am J Gastroenterol
3. Kuroki F, Iida M, Tominaga M, et al. Multiple vitamin status in Crohn's disease
: Correlation with disease activity. Dig Dis Sci
4. Imes S, Pinchbeck B, Dinwoodie A, Walker K, Thomson AB. Vitamin A
status in 137 patients with Crohn's disease
5. Kleinman RE, Balistreri WF, Heyman MB, et al. Nutritional support for patients with inflammatory bowel disease. J Pediatr Gastroenterol Nutr
6. Dean AD, Dean JA, Burton AH, Ricker RC. EPI-INFO: A word processing, database, and statistics program for epidemiology on microcomputers
. Stone Mountain, GA: USD, 1990.
7. Must A, Dallal GE, Dietz WH. Reference data for obesity: 85th and 95th percentiles of body mass index and triceps skinfold thickness.Am J Clin Nutr
8. Hyams JS, Ferry GD, Mandel FS, et al. Development and validation of a pediatric Crohn's disease
activity index. J Pediatr Gastroenterol Nutr
9. Kozarek RA, Patterson DJ, Gelfand MD, Botoman VA, Ball TJ, Wilske KR. Methotrexate induces clinical and histologic remission in patients with refractory inflammatory bowel disease. Ann Intern Med
10. Catignani GL, Bieri JG. Simultaneous determination of retinol
in serum or plasma by liquid chromatography.Clin Chem
11. Lockitch G, Halsted AC, Wadsworth L, Quigley G, Reston L, Jacobson B. Age- and sex-specific reference intervals and correlations for zinc, copper, selenium, iron, vitamins A and E, and related proteins.Clin Chem
12. Everitt BS. The analysis of contingency tables
. 2nd ed. London: Chapman and Hall, 1992:11-8.
13. Hosmer D, Lemeshow S. Applied logistic regression
. New York: John Wiley and Sons, 1989:25-44.
14. Schoelmerich MS, Becher MS, Hoppe-Seyler, et al. Zinc and vitamin A
deficiency in patients with Crohn's disease
is correlated with activity but not with localization or extent of the disease.Hepatogastroenterology
15. Shenai JP, Kennedy KA, Chytil F, Stahlman MT. Clinical trial of vitamin A
supplementation in infants susceptible to bronchopulmonary dysplasia. J Pediatr
16. Gadomaski AM, Kjothede CL, Wittpenn J, Bulux J, Rosas A, Forman MR. Conjunctival impression cytology to detect subclinical vitamin A
deficiency: Comparison of CIC with biochemical assessments. Am J Clin Nutr
17. Amedee-Manesme O, Luzeau R, Wittepen JR, Hanck A, Sommer A. Impression cytology detects subclinical vitamin A
deficiency. Am J Clin Nutr
18. Campos FA, Flores H, Underwood BA. Effect of an infection on vitamin A
status of children as measured by the relative dose response (RDR). Am J Clin Nutr
19. Varghese R, Ramakrishna BS, Jeyakumar S, Mathan M, Balasubramanian KA. Plasma vitamin A
is reduced in active ulcerative colitis
20. Grisham M, Granger DN. Neutrophil-mediated mucosal injury-role of reactive oxygen metabolites. Dig Dis Sci
21. Rousseau EJ, Davison AJ, Dunn B. Protection by beta-carotene and related compounds against oxygen-mediated cytotoxicity and genotoxicity: Implications for carcinogenesis and anticarcinogenesis.Free Radic Biol Med
22. Farrell MP, Roberts RJ. Vitamin E
. In: Shils ME, Olson JA, Shike M, eds. Modern nutrition in health and disease
. Malvern, PA: Lee and Febiger, 1994:326-41.
23. Norrby S, Sjodahl R, Tagesson C. Ineffectiveness of vitamin A
therapy in severe Crohn's disease
. Acta Chir Scand
24. Salazar-Lindo E, Salazar M, Alvarez JO. Association of diarrhea and low serum retinol
in Peruvian children. Am J Clin Nutr
25. Duggan C, Colin A, Higgins L, Rifai N. Vitamin A
status in acute exacerbations of cystic fibrosis. Am J Clin Nutr
26. Bhaskaram P, Reddy V, Raj S, Bhatnazar RC. Effect of measles on the nutritional status of preschool children. J Trop Med Hyg
The reader is also referred to another paper on this topic, published while this paper was being reviewed: Hoffenberg EJ, et al. Am J Clin Nutr 1997;65:1482-8.
Clinical Quiz Section
Short cases with accompanying photographs of diagnostic biopsies, endoscopic findings, or physical characteristics are welcomed for the newly established Clinical Quiz section. Please send a short history, glossy photograph, and answers to quiz questions to:
ESPGHAN Clinical Quiz Editor: Dr. Riccardo Troncone; Department of Pediatrics; University Federico II; Via S. Pansini, 5; 80131 - Naples; ITALY; Tel: 39-81-7463394; Fax: 39-81-5469811
NASPGN Clinical Quiz Editor: Dr. Joseph F. Fitzgerald; Department of Pediatrics; James Whitcomb Riley Hospital for Children; 702 Barnhill Drive; Indianapolis, IN 46202-5225; Tel: 317-274-3774; Fax: 317-274-8521