- Email: [email protected]
Lumbar Spine Bone Mineral Density at Diagnosis and During Follow-Up in Children With IBD
Journal of Clinical Densitometry, vol. 7, no. 3, 290–295, 2004 © Copyright 2004 by Humana Press Inc. All rights of any nature whatsoever reserved. 1094-6950/04/7:290–295/$25.00
Original Article
Lumbar Spine Bone Mineral Density at Diagnosis and During Follow-Up in Children With IBD Arun Gupta,1 Shirley Paski,1 Robert Issenman,1 and Colin Webber*,2 1McMaster
Children’s Hospital and Department of Pediatrics, McMaster University, Hamilton, Ontario, Canada and 2Department of Nuclear Medicine, Hamilton Health Sciences, Hamilton, Ontario, Canada
Abstract Lumbar spine body mineral density (BMD) was measured in 123 children (65 male, 58 female) suffering from inflammatory bowel disease (IBD) (82 Crohn’s disease, 41 ulcerative colitis) and in 46 children (25 male, 21 female) without any history of bone disease. Results in normal children showed that densitometer-derived reference values overestimated spine BMD, particularly for young children, such that the reported mean Z-scores for normal 10-yr-old children were –0.83 for males and –0.72 for females. For children with Crohn’s disease, the lumbar spine BMD was further reduced (Z-score = –1.44 for males, Z-score = –1.37 for females). For children with ulcerative colitis, the lumbar spine BMD was similar to that of normal children (Z-score = –0.93 for males, Z-score = –0.56 for females). There was no statistically significant reduction in average spine BMD Z-scores during follow-up periods ranging from 1.7 to 8.7 yr. When growth patterns were examined in individual children, six patients (three Crohn’s disease, three ulcerative colitis) were identified as losing spine BMD with respect to their baseline value and their expected pattern of BMD increase associated with normal growth. The children suffering from IBD who, most likely, will not maintain expected growth-related increases in spine BMD are those who are male, relatively young at diagnosis, and unlikely to be taking immunosuppressants. Key Words: Inflammatory bowel disease; Crohn’s disease; ulcerative colitis; lumbar spine bone mineral density; immunosuppressants.
Introduction
whole body was lower than healthy controls in 60 patients with Crohn’s disease, particularly those who had used corticosteroids. For patients with ulcerative colitis, BMD was similar to controls (2). Measurements of spine, hip, and total-body BMD within 6 mo of diagnosis in 68 patients matched with controls for age and gender showed no differences between patients and controls or between patients with Crohn’s disease and patients with ulcerative colitis (3). It was concluded that the subsequent development of osteoporosis was related to the disease process and/or the treatment of the disease. It is likely that the presence of IBD in children might not only result in loss of bone but might prevent the skeleton achieving the peak bone mass expected genetically. In children between 4 and 18 yr of age and suffering with IBD, spine and whole-body BMD was significantly reduced, especially for those with Crohn’s disease. Measurements repeated 2 yr later showed that
The prevalence of osteopenia and osteoporosis in patients with inflammatory bowel disease (IBD) has increased. For patients diagnosed with Crohn’s disease, spine and radius bone mineral density (BMD) has been shown to be low at baseline (1). In contrast, newly diagnosed patients with ulcerative colitis demonstrated no reduction in spine or radius BMD at baseline. In another study of a large group of patients with long-standing IBD, BMD of the spine, proximal femur, and
Received 12/09/03; Revised 03/02/04; Accepted 03/02/04. * Address correspondence to Colin Webber, Department of Nuclear Medicine, Hamilton Health Sciences, 1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5. E-mail: [email protected]
290
Spine BMD in Children With IBD the cumulative dose of prednisolone between measurements correlated negatively with the change in spine and body BMD (4). In five patients 10.6–16.8 yr of age and suffering from Crohn’s disease, vertebral compression fractures were evident. Each patient was on corticosteroids and spine BMD Z-scores ranged from –2.3 to –5.1 (5). There seems to be no doubt that, at least in women, the risk of fracture is increased in the presence of Crohn’s disease (6). The purpose of our study was to examine the deviation of lumbar spine BMD from that expected at diagnosis of IBD in a large number of children and to follow changes in spine BMD during subsequent treatment and follow-up of each child. In routine practice, measured BMD is categorized as normal or abnormal by comparison with expected for age-normal values derived from the bone densitometer reporting software. There is a tacit assumption that normal population values extracted from commercial densitometers are appropriate for the patient population studied. If the normal population values are marginally greater than is appropriate for the patient population, negative Z-scores will be reported even in the absence of any disease effect. If the normal population values are marginally less than is appropriate for the patient population, Z-scores might appear to be normal despite the presence of a bone mineral deficit. Consequently, in this study, we measured the spine BMD in a large number of children with IBD and in local, normal children. In the IBD patients, we then followed the changes in spine BMD over a number of years to examine the impact of disease progression and treatment upon the temporal changes in spine BMD.
Materials and Methods The lumbar spine BMD was measured on at least three separate occasions in each of 123 children with IBD using a Hologic 4500A densitometer (Hologic Inc., Bedford, MA, USA). All children were investigated with colonoscopy and biopsy and/or barium enema as well as small bowel follow through to yield a diagnosis of Crohn’s disease or ulcerative colitis according to accepted diagnostic criteria. There was no increased prevalence of fracture in these children at baseline or an increased incidence of fracture during follow-up. Children were followed clinically for parathyroid hormone (PTH), vitamin D, and Ca status and were treated if any deficiencies were identified. Eighty-two children were suffering from Crohn’s disease (43 male, 39 female) and 41 were suffering from ulcerative colitis (22 male, 19 female). The average age at the time of the first dual-energy X-ray absorptiometry (DXA) scan for children with Crohn’s was 11.8 ± 2.9 yr for males and 11.9 ± 2.4 yr for females. The corresponding ages for children with ulcerative colitis were 10.1 ± 2.8 yr and 11.7 ± 2.6 yr, respectively. The number of BMD measurements for each child varied from a minimum of three to a maximum of nine. The total duration of follow-up was 536 patient-years. The BMD was measured in a convenience sample of 25 normal males (age: 3.3–30.6 yr) and 21 normal females (age: 3.2–22.3 yr) recruited locally. The patients with IBD and the local, normal subjects were all Caucasian. Normal children completed a short questionnaire designed to identify those Journal of Clinical Densitometry
291 who were excessively active or were inactive, had body mass indexes outside the 95% expected range, consumed unusual diets, or had bone disease and/or a history of fractures. No child was excluded based on the responses to the questionnaire. Approval for the study was obtained from the McMaster University Research Ethics Board. Z-Scores for both patients with IBD and for local, normal subjects were taken from the Hologic database. Because the rate of bone mineral accrual in children changes dramatically with age, it is plausible that any deficit in lumbar spine BMD at baseline might be dependent on age and on the disease process of concern. Consequently, patient baseline Z-scores were examined as a function of age at baseline. Subsets of patients with Crohn’s disease (n = 16) and ulcerative colitis (n = 9) each had at least four BMD measurements over a minimum duration of 3.4 yr. Results for each child were fitted to a second-order polynomial in order to avoid any assumptions concerning patterns of growth in children with Crohn’s disease or ulcerative colitis while providing reasonable representations of the data for periods extending up to 8 yr in duration. We considered that a simple linear equation would not adequately represent bone growth, whereas equations such as exponentials or higher-order polynomials would impose unjustified assumptions on the patterns of growth or overrepresent the data in an unrealistic way. The overall goodness of fit for the 25 polynomials was expressed as the average deviation of each of the 160 measurements of spine BMD from the predicted BMD given by the polynomial. The average deviation was 0.020 g/cm2 or 2.6% when expressed as a percentage of the measured BMD. This percentage deviation was lower in older children as spine BMD increased. When the age-related patterns of change in spine BMD described by the fitted polynomials were examined, it was possible to categorize the 25 children into those who, over the course of follow-up, exhibited an increase in spine BMD (N = 6; Z-score increase > 0.75) and those who lost spine BMD (N = 6; Z-score decrease > 0.75). The Z-score thresholds for division into groups who gained or lost bone were selected arbitrarily to provide groups corresponding to the highest and lowest quartiles of age related rates of change of the Z-score. The two groups were contrasted in an attempt to identify features that might predict those children most likely to lose bone. The statistical significance of differences between groups was assessed using Student’s t-test. Significance was accepted when p < 0.05. It was assumed that a trend was present when 0.10 > p > 0.05.
Results Forty-two of 46 (91.3%) lumbar spine BMD measurements for the normal subjects fell within the 95% confidence interval for the normal population according to the densitometer database. For younger children, a greater proportion of results were below the expected-for-age curve. Figure 1 shows the densitometer-derived Z-scores for the normal subjects expressed as a function of age. Linear regression shows, for both male and female children, that the densitometer reference Volume 7, 2004
292
Gupta et al.
Fig. 1. Z-Scores calculated from the densitometer normal range as a function of age for normal males (●) and females (●) recruited locally. The lines represent the linear regression fits to the data for males (– – –) and females (—).
database was not representative of all local, normal children. The Z-score at age 5 averages about –1 for both males and females. For 20- to 25-yr-old normal subjects, the Z-score averages close to 0. Figure 2 shows lumbar spine BMD measurements for children suffering from Crohn’s disease. Spine BMD was generally below the densitometer expected-for-age values. The baseline Z-score (± standard deviation) for male children of all ages with Crohn’s disease was –1.44 ± 0.97, and for female children, it was –1.37 ± 1.22. The male and female Z-scores are not statistically different from each other. Lumbar spine BMD values for children with ulcerative colitis are shown in Fig. 3. Again, the spine BMD is, in general, less than the densitometer expectedfor-age value. Baseline Z-scores are –0.93 ± 1.10 for males and –0.56 ± 0.89 for females. These baseline Z-scores are not significantly different from each other. There is a trend for males with Crohn’s disease to have a lower BMD at baseline than males with ulcerative colitis (0.10 > p > 0.05). Females with Crohn’s disease had significantly lower spine BMDs at baseline than females with ulcerative colitis (p < 0.01). Figure 4 shows the baseline Z-scores for all children with IBD as a function of age at baseline. Comparison of the results for Crohn’s disease children (upper panel) with those shown in Fig. 1 for normal children suggest that, for younger children at baseline, there is no difference between Z-scores for normal children and those recently diagnosed with Crohn’s disease. The same is true for younger children with ulcerative colitis at baseline (Fig. 4, lower panel). For older children however, Fig. 1 shows that the baseline Z-score for normal 15-yr-old children should be about –0.5. Children with ulcerative colitis have such Z-scores at baseline, whereas those 15-yr-olds with recently diagnosed Crohn’s disease have Z-scores closer to –2. Table 1 compares baseline Z-scores with end-of-follow-up Zscores. The duration of follow-up for patients with Crohn’s disJournal of Clinical Densitometry
Fig. 2. Sequential measurements of lumbar spine BMD in 43 males (upper) and 39 females (lower) with Crohn’s disease. The solid continuous line gives the densitometer defined, mean spine BMD expected for age, and the two dashed lines give the densitometer defined, 95% confidence limits for normal children.
ease was 4.3 ± 1.9 yr for males and 4.1 ± 1.5 yr for females. For children with ulcerative colitis, the duration of follow-up was 4.9 ± 1.7 yr for males and 4.2 ± 1.7 yr for females. For all four groups, the mean end-of-follow-up Z-score was more negative than the mean baseline Z-score. However, when the Z-score change for each patient was averaged, there was no statistically significant reduction in spine BMD during the follow-up period. The upper panel of Fig. 5 shows the patterns of change in lumber spine BMD for nine boys with Crohn’s disease who were followed for periods ranging from 3.4 to 8.7 yr. Three patients (shown as dashed lines) lost considerable spine BMD relative to their own baseline BMD and their expected increase in BMD during growth. Two children could be classified into the bonegain group (heavy solid lines). The lower panel of Fig. 5 shows patterns of change for six girls with Crohn’s disease who were followed from 5.2 to 8.3 yr. None of the girls lost BMD with respect to their baseline BMD and expected growth-related increases and two could be classified as having gained bone. The upper panel of Fig. 6 shows patterns of change for lumber spine BMD in four boys with ulcerative colitis who were Volume 7, 2004
Spine BMD in Children With IBD
Fig. 3. Sequential measurements of lumbar spine BMD in 25 males (upper) and 19 females (lower) with ulcerative colitis. The normal range is as defined in Fig. 2.
followed for periods ranging from 6.4 to 8.6 yr. Three of the four patients lost spine BMD relative to their own baseline BMD and their expected BMD increase during growth; one was considered to have gained BMD. The lower panel of Fig. 6 shows the patterns of change for five girls with ulcerative colitis who were followed from 5.7 to 7.9 yr. None lost BMD with respect to their baseline BMD and expected growth; one subject was classified as having gained BMD.
Discussion At the time of diagnosis of Crohn’s disease, the lumbar spine BMD Z-score for the children in this study averaged 1.39 standard deviations below the value expected for age. For the children with ulcerative cholitis, the corresponding figure was 0.76 below expected. At first sight, this would suggest that most children with IBD would be expected to have a low spine BMD at diagnosis. However, our measurements in normal children recruited locally show that a large fraction of this reduced Z-score is, in fact, the result of the use of normal values that are inappropriate for children of southern Ontario. This observation serves to emphasize the importance of the verification of manufacturer-supplied normal values. The normal population described by the values provided with the Journal of Clinical Densitometry
293
Fig. 4. Baseline Z-scores as a function of age in male (●) and female (●) children with Crohn’s disease (upper) and ulcerative colitis (lower). The corresponding data for normal children are shown in Fig. 1.
Hologic 4500A appear to yield higher than expected agedependent values for normal, younger children within our local population. This might be explained by ethnic differences between the reference and local populations or by differences in lifestyle factors that influence children’s BMD. It might be that differences between baseline BMD measurements in various studies of children with IBD reflect differences in reference populations rather than true differences in patient populations. Without doubt, spine BMD was lower at diagnosis for children with Crohn’s disease than for children with ulcerative colitis. This was true for both male (0.10 > p > 0.05) and female (p < 0.01) children and might reflect simply a chronic impact of Crohn’s disease on the skeleton compared to the acute influence of ulcerative colitis. However, it is likely that mineral metabolism is disturbed in Crohn’s disease because it is known that Crohn’s disease is associated with disorganized mineral and osteoid and that osteoblasts are adversely affected (7). The Z-scores at end of follow-up were more negative than baseline Z-scores for both male and female children with either Crohn’s disease or ulcerative colitis, although none of the further Volume 7, 2004
294
Gupta et al. Table 1 Z-Scores for Lumbar Spine BMD at Baseline and End of Follow-Up for Children With Crohn’s Disease or Ulcerative Colitis (UC)
Group Crohn’s male (N = 43) Crohn’s female (N = 39) UC male (N = 22) UC female (N = 19)
Group Z-score (±SD) at baseline
Group Z-score (±SD) at end of follow-up
Mean (±SD) of individual differences
–1.44 ± 0.97
–1.68 ± 1.16
–0.26 ± 1.10
–1.37 ± 1.22
–1.38 ± 1.07
–0.01 ± 0.89
–0.93 ± 1.10
–1.29 ± 1.40
–0.35 ± 1.19
–0.56 ± 0.89
–0.62 ± 0.99
0.00 ± 0.86
Fig. 6. Long-term follow-up of lumbar spine BMD for four males (upper) and five females (bottom) with ulcerative colitis. The normal range is as defined in Fig. 2. The patterns of BMD gain are as described in Fig. 5.
Fig. 5. Long-term follow-up of lumbar spine BMD for nine males (upper) and seven females (lower) with Crohn’s disease. The normal range is as defined in Fig. 2. The thick dashed lines represent children who lost spine BMD with respect to their own baseline and expected pattern of growth. The thick solid lines represent children who gained spine BMD with respect to their own baseline and expected pattern of growth. Journal of Clinical Densitometry
BMD losses was statistically significant. Of the 25 children in whom sufficient longitudinal data were available to justify examination of age-dependent patterns of change in spine BMD, six showed progressive spine BMD reduction with respect to their baseline BMD and their expected growth pattern. No single factor could uniquely identify these six children. All were male, but not all males lost bone. No female child with IBD showed a relative loss of spine BMD. Three of the six BMD losers suffered from Crohn’s disease and three suffered from ulcerative colitis. Generally, the six patients had their baseline measurements of spine BMD at a young age (Crohn’s disease, 9.3 ± 0.9 yr; ulcerative colitis, 7.8 ± 1.9 yr). One of the six was on corticosteroids and one was on both corticosteroids and azathioprine. The remaining four (67%) received no immunosuppressants. Of the remaining 19 children, 13 had Crohn’s disease and 6 had ulcerative colitis. Eight were on corticosteroids, seven were on both corticosteroids and azathioprine, and only two (11%) received no immunosuppressants. These results suggest that generally children suffering from IBD who will not sustain a normal pattern of increase in spine BMD during growth will be those males diagnosed at an early age who receive no immunosuppressants. Volume 7, 2004
Spine BMD in Children With IBD In conclusion, this work has underlined the importance of each bone densitometry laboratory establishing normal values for all segments of the local patient population. In addition, our results suggest that patients with IBD who are unlikely to achieve normal patterns of growth will be young males not on immunosuppressants.
References 1. Ghosh S, Cowen S, Hannon WJ, Ferguson A. 1994 Low bone mineral density in Crohn’s disease but not in ulcerative cholitis at diagnosis. Gastroenterology 107:1031–1039. 2. Jahnsen J, Falch JA, Aadland E, Mowinckel P. 1997 Bone mineral density is reduced in patients with Crohn’s disease but not in patients with ulcerative colitis: a population based study. Gut 40:313–319.
Journal of Clinical Densitometry
295 3. Schoon EJ, Blok BM, Geerling BJ, Russel MG, Stockbrügger RW, Brummer R-J M. 2000 Bone mineral density in patients with recently diagnosed inflammatory bowel disease. Gastroenterology 119:1203–1208. 4. Boot AM, Bouquet J, Krenning EP, de Muinck Keizer-Schrama SMPF. 1998 Bone mineral density and nutritional status in children with chronic inflammatory bowel disease. Gut 42:188–194. 5. Semeao EJ, Stallings VA, Peck SN, Piccoli DA. 1997 Vertebral compression fractures in pediatric patients with Crohn’s disease. Gastroenterology 112:1710–1713. 6. Arden NK, Cooper C. 2002 Osteoporosis in patients with inflammatory bowel disease. Gut 50:9–10. 7. Hyams JS, Wyzga N, Kreutzer DL, Justinich CJ, Gronowicz GA. 1997 Alterations in bone metabolism in children with inflammatory bowel disease: an in vitro study. J Pediatr Gastroenterol Nutr 24:289–295.
Volume 7, 2004
Original Article
Lumbar Spine Bone Mineral Density at Diagnosis and During Follow-Up in Children With IBD Arun Gupta,1 Shirley Paski,1 Robert Issenman,1 and Colin Webber*,2 1McMaster
Children’s Hospital and Department of Pediatrics, McMaster University, Hamilton, Ontario, Canada and 2Department of Nuclear Medicine, Hamilton Health Sciences, Hamilton, Ontario, Canada
Abstract Lumbar spine body mineral density (BMD) was measured in 123 children (65 male, 58 female) suffering from inflammatory bowel disease (IBD) (82 Crohn’s disease, 41 ulcerative colitis) and in 46 children (25 male, 21 female) without any history of bone disease. Results in normal children showed that densitometer-derived reference values overestimated spine BMD, particularly for young children, such that the reported mean Z-scores for normal 10-yr-old children were –0.83 for males and –0.72 for females. For children with Crohn’s disease, the lumbar spine BMD was further reduced (Z-score = –1.44 for males, Z-score = –1.37 for females). For children with ulcerative colitis, the lumbar spine BMD was similar to that of normal children (Z-score = –0.93 for males, Z-score = –0.56 for females). There was no statistically significant reduction in average spine BMD Z-scores during follow-up periods ranging from 1.7 to 8.7 yr. When growth patterns were examined in individual children, six patients (three Crohn’s disease, three ulcerative colitis) were identified as losing spine BMD with respect to their baseline value and their expected pattern of BMD increase associated with normal growth. The children suffering from IBD who, most likely, will not maintain expected growth-related increases in spine BMD are those who are male, relatively young at diagnosis, and unlikely to be taking immunosuppressants. Key Words: Inflammatory bowel disease; Crohn’s disease; ulcerative colitis; lumbar spine bone mineral density; immunosuppressants.
Introduction
whole body was lower than healthy controls in 60 patients with Crohn’s disease, particularly those who had used corticosteroids. For patients with ulcerative colitis, BMD was similar to controls (2). Measurements of spine, hip, and total-body BMD within 6 mo of diagnosis in 68 patients matched with controls for age and gender showed no differences between patients and controls or between patients with Crohn’s disease and patients with ulcerative colitis (3). It was concluded that the subsequent development of osteoporosis was related to the disease process and/or the treatment of the disease. It is likely that the presence of IBD in children might not only result in loss of bone but might prevent the skeleton achieving the peak bone mass expected genetically. In children between 4 and 18 yr of age and suffering with IBD, spine and whole-body BMD was significantly reduced, especially for those with Crohn’s disease. Measurements repeated 2 yr later showed that
The prevalence of osteopenia and osteoporosis in patients with inflammatory bowel disease (IBD) has increased. For patients diagnosed with Crohn’s disease, spine and radius bone mineral density (BMD) has been shown to be low at baseline (1). In contrast, newly diagnosed patients with ulcerative colitis demonstrated no reduction in spine or radius BMD at baseline. In another study of a large group of patients with long-standing IBD, BMD of the spine, proximal femur, and
Received 12/09/03; Revised 03/02/04; Accepted 03/02/04. * Address correspondence to Colin Webber, Department of Nuclear Medicine, Hamilton Health Sciences, 1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5. E-mail: [email protected]
290
Spine BMD in Children With IBD the cumulative dose of prednisolone between measurements correlated negatively with the change in spine and body BMD (4). In five patients 10.6–16.8 yr of age and suffering from Crohn’s disease, vertebral compression fractures were evident. Each patient was on corticosteroids and spine BMD Z-scores ranged from –2.3 to –5.1 (5). There seems to be no doubt that, at least in women, the risk of fracture is increased in the presence of Crohn’s disease (6). The purpose of our study was to examine the deviation of lumbar spine BMD from that expected at diagnosis of IBD in a large number of children and to follow changes in spine BMD during subsequent treatment and follow-up of each child. In routine practice, measured BMD is categorized as normal or abnormal by comparison with expected for age-normal values derived from the bone densitometer reporting software. There is a tacit assumption that normal population values extracted from commercial densitometers are appropriate for the patient population studied. If the normal population values are marginally greater than is appropriate for the patient population, negative Z-scores will be reported even in the absence of any disease effect. If the normal population values are marginally less than is appropriate for the patient population, Z-scores might appear to be normal despite the presence of a bone mineral deficit. Consequently, in this study, we measured the spine BMD in a large number of children with IBD and in local, normal children. In the IBD patients, we then followed the changes in spine BMD over a number of years to examine the impact of disease progression and treatment upon the temporal changes in spine BMD.
Materials and Methods The lumbar spine BMD was measured on at least three separate occasions in each of 123 children with IBD using a Hologic 4500A densitometer (Hologic Inc., Bedford, MA, USA). All children were investigated with colonoscopy and biopsy and/or barium enema as well as small bowel follow through to yield a diagnosis of Crohn’s disease or ulcerative colitis according to accepted diagnostic criteria. There was no increased prevalence of fracture in these children at baseline or an increased incidence of fracture during follow-up. Children were followed clinically for parathyroid hormone (PTH), vitamin D, and Ca status and were treated if any deficiencies were identified. Eighty-two children were suffering from Crohn’s disease (43 male, 39 female) and 41 were suffering from ulcerative colitis (22 male, 19 female). The average age at the time of the first dual-energy X-ray absorptiometry (DXA) scan for children with Crohn’s was 11.8 ± 2.9 yr for males and 11.9 ± 2.4 yr for females. The corresponding ages for children with ulcerative colitis were 10.1 ± 2.8 yr and 11.7 ± 2.6 yr, respectively. The number of BMD measurements for each child varied from a minimum of three to a maximum of nine. The total duration of follow-up was 536 patient-years. The BMD was measured in a convenience sample of 25 normal males (age: 3.3–30.6 yr) and 21 normal females (age: 3.2–22.3 yr) recruited locally. The patients with IBD and the local, normal subjects were all Caucasian. Normal children completed a short questionnaire designed to identify those Journal of Clinical Densitometry
291 who were excessively active or were inactive, had body mass indexes outside the 95% expected range, consumed unusual diets, or had bone disease and/or a history of fractures. No child was excluded based on the responses to the questionnaire. Approval for the study was obtained from the McMaster University Research Ethics Board. Z-Scores for both patients with IBD and for local, normal subjects were taken from the Hologic database. Because the rate of bone mineral accrual in children changes dramatically with age, it is plausible that any deficit in lumbar spine BMD at baseline might be dependent on age and on the disease process of concern. Consequently, patient baseline Z-scores were examined as a function of age at baseline. Subsets of patients with Crohn’s disease (n = 16) and ulcerative colitis (n = 9) each had at least four BMD measurements over a minimum duration of 3.4 yr. Results for each child were fitted to a second-order polynomial in order to avoid any assumptions concerning patterns of growth in children with Crohn’s disease or ulcerative colitis while providing reasonable representations of the data for periods extending up to 8 yr in duration. We considered that a simple linear equation would not adequately represent bone growth, whereas equations such as exponentials or higher-order polynomials would impose unjustified assumptions on the patterns of growth or overrepresent the data in an unrealistic way. The overall goodness of fit for the 25 polynomials was expressed as the average deviation of each of the 160 measurements of spine BMD from the predicted BMD given by the polynomial. The average deviation was 0.020 g/cm2 or 2.6% when expressed as a percentage of the measured BMD. This percentage deviation was lower in older children as spine BMD increased. When the age-related patterns of change in spine BMD described by the fitted polynomials were examined, it was possible to categorize the 25 children into those who, over the course of follow-up, exhibited an increase in spine BMD (N = 6; Z-score increase > 0.75) and those who lost spine BMD (N = 6; Z-score decrease > 0.75). The Z-score thresholds for division into groups who gained or lost bone were selected arbitrarily to provide groups corresponding to the highest and lowest quartiles of age related rates of change of the Z-score. The two groups were contrasted in an attempt to identify features that might predict those children most likely to lose bone. The statistical significance of differences between groups was assessed using Student’s t-test. Significance was accepted when p < 0.05. It was assumed that a trend was present when 0.10 > p > 0.05.
Results Forty-two of 46 (91.3%) lumbar spine BMD measurements for the normal subjects fell within the 95% confidence interval for the normal population according to the densitometer database. For younger children, a greater proportion of results were below the expected-for-age curve. Figure 1 shows the densitometer-derived Z-scores for the normal subjects expressed as a function of age. Linear regression shows, for both male and female children, that the densitometer reference Volume 7, 2004
292
Gupta et al.
Fig. 1. Z-Scores calculated from the densitometer normal range as a function of age for normal males (●) and females (●) recruited locally. The lines represent the linear regression fits to the data for males (– – –) and females (—).
database was not representative of all local, normal children. The Z-score at age 5 averages about –1 for both males and females. For 20- to 25-yr-old normal subjects, the Z-score averages close to 0. Figure 2 shows lumbar spine BMD measurements for children suffering from Crohn’s disease. Spine BMD was generally below the densitometer expected-for-age values. The baseline Z-score (± standard deviation) for male children of all ages with Crohn’s disease was –1.44 ± 0.97, and for female children, it was –1.37 ± 1.22. The male and female Z-scores are not statistically different from each other. Lumbar spine BMD values for children with ulcerative colitis are shown in Fig. 3. Again, the spine BMD is, in general, less than the densitometer expectedfor-age value. Baseline Z-scores are –0.93 ± 1.10 for males and –0.56 ± 0.89 for females. These baseline Z-scores are not significantly different from each other. There is a trend for males with Crohn’s disease to have a lower BMD at baseline than males with ulcerative colitis (0.10 > p > 0.05). Females with Crohn’s disease had significantly lower spine BMDs at baseline than females with ulcerative colitis (p < 0.01). Figure 4 shows the baseline Z-scores for all children with IBD as a function of age at baseline. Comparison of the results for Crohn’s disease children (upper panel) with those shown in Fig. 1 for normal children suggest that, for younger children at baseline, there is no difference between Z-scores for normal children and those recently diagnosed with Crohn’s disease. The same is true for younger children with ulcerative colitis at baseline (Fig. 4, lower panel). For older children however, Fig. 1 shows that the baseline Z-score for normal 15-yr-old children should be about –0.5. Children with ulcerative colitis have such Z-scores at baseline, whereas those 15-yr-olds with recently diagnosed Crohn’s disease have Z-scores closer to –2. Table 1 compares baseline Z-scores with end-of-follow-up Zscores. The duration of follow-up for patients with Crohn’s disJournal of Clinical Densitometry
Fig. 2. Sequential measurements of lumbar spine BMD in 43 males (upper) and 39 females (lower) with Crohn’s disease. The solid continuous line gives the densitometer defined, mean spine BMD expected for age, and the two dashed lines give the densitometer defined, 95% confidence limits for normal children.
ease was 4.3 ± 1.9 yr for males and 4.1 ± 1.5 yr for females. For children with ulcerative colitis, the duration of follow-up was 4.9 ± 1.7 yr for males and 4.2 ± 1.7 yr for females. For all four groups, the mean end-of-follow-up Z-score was more negative than the mean baseline Z-score. However, when the Z-score change for each patient was averaged, there was no statistically significant reduction in spine BMD during the follow-up period. The upper panel of Fig. 5 shows the patterns of change in lumber spine BMD for nine boys with Crohn’s disease who were followed for periods ranging from 3.4 to 8.7 yr. Three patients (shown as dashed lines) lost considerable spine BMD relative to their own baseline BMD and their expected increase in BMD during growth. Two children could be classified into the bonegain group (heavy solid lines). The lower panel of Fig. 5 shows patterns of change for six girls with Crohn’s disease who were followed from 5.2 to 8.3 yr. None of the girls lost BMD with respect to their baseline BMD and expected growth-related increases and two could be classified as having gained bone. The upper panel of Fig. 6 shows patterns of change for lumber spine BMD in four boys with ulcerative colitis who were Volume 7, 2004
Spine BMD in Children With IBD
Fig. 3. Sequential measurements of lumbar spine BMD in 25 males (upper) and 19 females (lower) with ulcerative colitis. The normal range is as defined in Fig. 2.
followed for periods ranging from 6.4 to 8.6 yr. Three of the four patients lost spine BMD relative to their own baseline BMD and their expected BMD increase during growth; one was considered to have gained BMD. The lower panel of Fig. 6 shows the patterns of change for five girls with ulcerative colitis who were followed from 5.7 to 7.9 yr. None lost BMD with respect to their baseline BMD and expected growth; one subject was classified as having gained BMD.
Discussion At the time of diagnosis of Crohn’s disease, the lumbar spine BMD Z-score for the children in this study averaged 1.39 standard deviations below the value expected for age. For the children with ulcerative cholitis, the corresponding figure was 0.76 below expected. At first sight, this would suggest that most children with IBD would be expected to have a low spine BMD at diagnosis. However, our measurements in normal children recruited locally show that a large fraction of this reduced Z-score is, in fact, the result of the use of normal values that are inappropriate for children of southern Ontario. This observation serves to emphasize the importance of the verification of manufacturer-supplied normal values. The normal population described by the values provided with the Journal of Clinical Densitometry
293
Fig. 4. Baseline Z-scores as a function of age in male (●) and female (●) children with Crohn’s disease (upper) and ulcerative colitis (lower). The corresponding data for normal children are shown in Fig. 1.
Hologic 4500A appear to yield higher than expected agedependent values for normal, younger children within our local population. This might be explained by ethnic differences between the reference and local populations or by differences in lifestyle factors that influence children’s BMD. It might be that differences between baseline BMD measurements in various studies of children with IBD reflect differences in reference populations rather than true differences in patient populations. Without doubt, spine BMD was lower at diagnosis for children with Crohn’s disease than for children with ulcerative colitis. This was true for both male (0.10 > p > 0.05) and female (p < 0.01) children and might reflect simply a chronic impact of Crohn’s disease on the skeleton compared to the acute influence of ulcerative colitis. However, it is likely that mineral metabolism is disturbed in Crohn’s disease because it is known that Crohn’s disease is associated with disorganized mineral and osteoid and that osteoblasts are adversely affected (7). The Z-scores at end of follow-up were more negative than baseline Z-scores for both male and female children with either Crohn’s disease or ulcerative colitis, although none of the further Volume 7, 2004
294
Gupta et al. Table 1 Z-Scores for Lumbar Spine BMD at Baseline and End of Follow-Up for Children With Crohn’s Disease or Ulcerative Colitis (UC)
Group Crohn’s male (N = 43) Crohn’s female (N = 39) UC male (N = 22) UC female (N = 19)
Group Z-score (±SD) at baseline
Group Z-score (±SD) at end of follow-up
Mean (±SD) of individual differences
–1.44 ± 0.97
–1.68 ± 1.16
–0.26 ± 1.10
–1.37 ± 1.22
–1.38 ± 1.07
–0.01 ± 0.89
–0.93 ± 1.10
–1.29 ± 1.40
–0.35 ± 1.19
–0.56 ± 0.89
–0.62 ± 0.99
0.00 ± 0.86
Fig. 6. Long-term follow-up of lumbar spine BMD for four males (upper) and five females (bottom) with ulcerative colitis. The normal range is as defined in Fig. 2. The patterns of BMD gain are as described in Fig. 5.
Fig. 5. Long-term follow-up of lumbar spine BMD for nine males (upper) and seven females (lower) with Crohn’s disease. The normal range is as defined in Fig. 2. The thick dashed lines represent children who lost spine BMD with respect to their own baseline and expected pattern of growth. The thick solid lines represent children who gained spine BMD with respect to their own baseline and expected pattern of growth. Journal of Clinical Densitometry
BMD losses was statistically significant. Of the 25 children in whom sufficient longitudinal data were available to justify examination of age-dependent patterns of change in spine BMD, six showed progressive spine BMD reduction with respect to their baseline BMD and their expected growth pattern. No single factor could uniquely identify these six children. All were male, but not all males lost bone. No female child with IBD showed a relative loss of spine BMD. Three of the six BMD losers suffered from Crohn’s disease and three suffered from ulcerative colitis. Generally, the six patients had their baseline measurements of spine BMD at a young age (Crohn’s disease, 9.3 ± 0.9 yr; ulcerative colitis, 7.8 ± 1.9 yr). One of the six was on corticosteroids and one was on both corticosteroids and azathioprine. The remaining four (67%) received no immunosuppressants. Of the remaining 19 children, 13 had Crohn’s disease and 6 had ulcerative colitis. Eight were on corticosteroids, seven were on both corticosteroids and azathioprine, and only two (11%) received no immunosuppressants. These results suggest that generally children suffering from IBD who will not sustain a normal pattern of increase in spine BMD during growth will be those males diagnosed at an early age who receive no immunosuppressants. Volume 7, 2004
Spine BMD in Children With IBD In conclusion, this work has underlined the importance of each bone densitometry laboratory establishing normal values for all segments of the local patient population. In addition, our results suggest that patients with IBD who are unlikely to achieve normal patterns of growth will be young males not on immunosuppressants.
References 1. Ghosh S, Cowen S, Hannon WJ, Ferguson A. 1994 Low bone mineral density in Crohn’s disease but not in ulcerative cholitis at diagnosis. Gastroenterology 107:1031–1039. 2. Jahnsen J, Falch JA, Aadland E, Mowinckel P. 1997 Bone mineral density is reduced in patients with Crohn’s disease but not in patients with ulcerative colitis: a population based study. Gut 40:313–319.
Journal of Clinical Densitometry
295 3. Schoon EJ, Blok BM, Geerling BJ, Russel MG, Stockbrügger RW, Brummer R-J M. 2000 Bone mineral density in patients with recently diagnosed inflammatory bowel disease. Gastroenterology 119:1203–1208. 4. Boot AM, Bouquet J, Krenning EP, de Muinck Keizer-Schrama SMPF. 1998 Bone mineral density and nutritional status in children with chronic inflammatory bowel disease. Gut 42:188–194. 5. Semeao EJ, Stallings VA, Peck SN, Piccoli DA. 1997 Vertebral compression fractures in pediatric patients with Crohn’s disease. Gastroenterology 112:1710–1713. 6. Arden NK, Cooper C. 2002 Osteoporosis in patients with inflammatory bowel disease. Gut 50:9–10. 7. Hyams JS, Wyzga N, Kreutzer DL, Justinich CJ, Gronowicz GA. 1997 Alterations in bone metabolism in children with inflammatory bowel disease: an in vitro study. J Pediatr Gastroenterol Nutr 24:289–295.
Volume 7, 2004