Professor G. Bianchi Porro, Cattedra di Gastroenterologia, Azienda Ospedaliera, Polo Universitario ‘L. Sacco’, Via G.B. Grassi 74, 20157 Milano, Italy (fax: + 39 2 35799232).
Abstract. Ardizzone S, Bollani S, Bettica P, Bevilacqua M, Molteni P, Bianchi Porro G (‘L. Sacco’ University-Hospital, Milan, Italy). Altered bone metabolism in inflammatory bowel disease: there is a difference between Crohn’s disease and ulcerative colitis. J Intern Med 2000; 247: 63–70.
Objectives. The aims of this study were to assess bone metabolism in inflammatory bowel disease (IBD) patients and to evaluate potential differences between Crohn’s disease (CD) and ulcerative colitis (UC) with respect to the mechanisms underlying bone loss in this group of diseases.
Design and setting. This was a cross-sectional study which started in 1992. Patients were randomly selected for invitation to participate and were examined during the years 1992–95 in one research clinic in Milan.
Subjects and methods. Fifty-one patients suffering from CD (30 women and 21 men, mean age 38.7 ± 13.2 years) and 40 with UC (15 women and 25 men, mean age 34.4. ± 12.5 years) entered the study. Thirty healthy subjects were selected as sex- and age-matched controls (C). Spine and femoral neck bone mineral density (expressed as T score), calciotropic hormones (parathyroid hormone, PTH; 25-hydroxycholecalciferol, 25(OH)D3; 1,25-hydroxycholecalciferol, 1,25(OH)D3) and biochemical markers of bone turnover (ostecalcin, OC; total alkaline phosphatase, ALP; type I collagen C-terminal telopeptide, ICTP) were evaluated.
Results. Spine and femur T scores were similar in the two groups (spine: CD = –1.49 ± 1.46; UC = –1.67 ± 1.13; femur: CD = –1.80 ± 1.36; UC = –1.60 ± 1.03). Based upon the WHO guidelines, only 8% of CD patients and 15% of UC patients had a normal bone mineral density (BMD), 55% (CD) and 67% (UC) were osteopenic, and 37% (CD) and 18% (UC) were osteoporotic. The distribution amongst the three different diagnostic groups was not significantly different between CD and UC groups (P = 0.11). PTH and 25(OH)D3 concentrations were not significantly different between CD and UC patients and controls, whilst 1,25(OH)D3 concentrations were significantly lower in both CD and UC patients compared with controls (P < 0.05). Bone turnover was increased in UC but not in CD patients, as shown by significantly increased concentrations in UC patients of both OC (CD = 7.77 ± 5.06, UC = 10.03 ± 6.24, C = 6.58 ± 2.87, P < 0.05 vs. C) and ICTP (CD = 5.74 ± 3.94, UC = 10.2 ± 8.47, C = 3.48 ± 0.95, P < 0.05 vs. CD and C). In a stepwise regression that included age, sex, disease duration and cumulative prednisolone dose as independent variables, the femur T score was significantly inversely related to disease duration (r2 = 0.125, F = 6.06) in CD patients. In UC patients, the spine T score was inversely related to age (r2 = 0.107, F = 5.49) and significantly related to sex (more negative in males: r2 = 0.3, F = 16.1); the femur T score was significantly related to sex (more negative in males) and inversely related to the cumulative prednisolone dose (r2 = 0.283, F = 7.3).
Conclusions. These data show that IBD patients have a diffuse osteopenia, the degree of which is not different in CD and UC; however, bone turnover is significantly higher in UC. Finally, osteopenia is related to disease duration in CD, whilst it is related to the male sex and glucocorticoid treatment in UC.
Low bone mineral density (BMD) has been demonstrated in patients with inflammatory bowel disease (IBD) [1–11]. The exact pathogenesis for this is not clearly understood, but it is likely to be multifactorial. Several conditions such as corticosteroid therapy, calcium and vitamin D deficiency, hypogonadism, malnutrition, smoking, alcohol consumption and reduced physical activity are all potential contributory factors. Moreover, it is well known that IBD patients show high circulation levels of cytokines  such as interleukin 1, tumour necrosis factor, transforming growth factor α, interleukin 6 and interleukin 4. These inflammatory factors stimulate the osteoclast function [13, 14], effector of bone resorption, and could inhibit osteoblast , the mediator of bone formation, with potential deleterious effects on bone density.
Several studies have been performed in an effort to establish which of these factors really acts in determining bone loss in IBD, often providing conflicting data and reflecting the heterogeneous nature of the patient cohorts included. Thus, in some of these studies, corticosteroid therapy appeared as the main osteopenic factor [1–4], whereas in others [5, 6] low BMD was found in the absence of past or present steroid treatment, although very low T scores were most often observed in patients receiving large doses . Furthermore, various other factors such as small intestinal resection , ileal involvement , body mass index (BMI) and height [4, 6, 9, 10] have been correlated with bone loss in IBD. These differences are probably the consequence not only of extreme variability in patient selection, but also of the evaluation methods, the sites studied (radial, lumbar spine and hip) and the expression of reported results which differ in their sensitivity and specificity. Moreover, most published studies show no comparative evaluation in BMD and biochemical markers of bone turnover between Crohn’s disease (CD) and ulcerative colitis (UC), and patients with CD and UC were regarded as belonging to one group.
The aim of this cross-sectional study was to assess bone metabolism by measuring BMD, calcium homeostasis and biochemical markers of bone turnover, comparing CD with UC patients and healthy controls. In addition, a special effort was made to evaluate differences between CD and UC with respect to the pathogenetic mechanisms underlying bone loss in this group of diseases.
Materials and methods
The study protocol was performed in accordance with the Declaration of Helsinki and was approved by the ethics committee of our hospital. Patients and controls gave written informed consent before the investigation.
Patients and controls
Eligible patients invited to participate in this study were men and women aged between 18 and 75 years attending our gastrointestinal unit who had a confirmed diagnosis of CD and UC . The diagnosis of CD and UC was carried out by a combination of radiology, endoscopy and histology. Exclusion criteria were conditions believed to be at risk for altered bone metabolism, such as cardiopulmonary, hepatic or renal disease, hypogonadism and thyrotoxicosis. Moreover, patients with inflammatory joint disease (ankylosing spondylitis, rheumatoid arthritis etc.), disease confined to the rectum, those who were pregnant, who were already receiving treatment for osteoporosis (bisphosphonates, calcium, vitamin D, fluoride, calcitonin and hormone replacement therapy), who had severe malabsorption and who were suffering from malnutrition were also excluded.
Clinical assessment included duration of disease, previous bowel resections, current steroid treatment and cumulative lifetime steroid dose (expressed in g of prednisone), extent of disease, clinical activity of bowel disease and BMI (weight/height2). Crohn’s Disease Activity Index (CDAI)  and Truelove and Witts’ score were used to monitor the activity of CD and UC, respectively. Remission was defined as a CDAI < 150 in CD patients, and as absence of symptoms of active disease and no signs of active inflammation on sigmoidoscopy in UC patients. All female patients were asked for their menstrual status.
The control group was matched for sex and age (± 2 years) against CD and UC patients, and included 30 healthy subjects who were randomly drawn from 130 healthy normal volunteers comprising our Italian reference population for bone mass measurements. None of them was taking any medication known to affect bone turnover and none had metabolic bone disease or recent fracture.
Both the patients and the controls represented several different occupational groups from manual workers to office personnel.
Measurement of bone mineral density
Bone mineral density (g cm–2) was measured by a single operator during 1 week using blood sample collections. Measurement was by dual-energy X-ray absorptiometry (DEXA) using a Hologic QDR2000 densitometer (Hologic, Waltham, MA). The densitometer was calibrated every day with the Hologic spine and left proximal femoral neck phantom, and the in vitro coefficient of variation was within 0.5% for the lumbar spine, the femoral neck, the Ward’s triangle and the trochanter. The measurement sites were the lumbar spine (L2–L4) and the femoral neck, the Ward’s triangle and the trochanter in the left proximal femur; an average (total femur) was obtained from the three sites.
The results are presented as either Z scores (number of standard deviations from normal values corrected for sex and age) or T scores, which are standard deviation scores from peak bone mass above or below the adjusted reference values obtained in healthy young adults.
Based upon the World Health Organization (WHO) guidelines , to determine the prevalence in patient groups of low BMD, the results were expressed as T scores. For this purpose, considering that at the proximal femur level both trabecular (Ward’s triangle) and cortical bones are represented, femur T scores are reported. In particular, the following diagnostic categories were used: (i) normal BMD, T score above –1; (ii) osteopenia, T score between –1 and –2.5; (iii) osteoporosis, T score below –2.5; (iv) severe or established osteoporosis, T score below –2.5 and the presence of one or more fragility fractures. However, it should be noted that these thresholds have only been developed in women and that appropriate diagnostic criteria in men have not been established.
The normal values were provided by the Hologic reference database obtained from healthy normal Caucasian volunteers of either sex. Because the Hologic reference database is based on a USA population, we recruited 130 healthy normal volunteers from the local Italian population. BMD values obtained from locally recruited control subjects were not significantly different from the Hologic reference database; hence, this database was used for our control values.
Biochemical assessment of bone metabolism
After overnight fasting, all blood samples were always collected in ice at 08.00–09.00 h. After centrifugation, serum and plasma were stored at –20°C until needed. All samples were measured using commercially available techniques normally used in the Department of Endocrinology and Medicine Laboratory of our university hospital.
Calcium and calciotropic hormones. Plasma calcium (normal range 8.5–10.5 mg dL–1) was measured by the o-cresolphtalein complexone method (Boehringer Mannheim Systems, Mannheim, Germany).
Serum parathyroid hormone level (PTH) was measured by an immunoradiometric assay (Allegro Intact IRMA PTH, Nichols, USA) that measures the intact molecule (normal range 10–70 pg mL–1). The intra-assay and interassay coefficients of variation for this assay are 3.8 and 6%, respectively.
Serum 25-hydroxycholecalciferol [25(OH)D3] level was measured by radioimmunoassay (Nichols Diagnostic Institute, USA; normal range 15–40 ng mL–1). The intra-assay and interassay coefficients of variation for this assay are 5.6 and 8.3%, respectively.
Serum 1,25-hydroxycholecalciferol [1,25(OH)D3] level was measured by a radioreceptor assay (Nichols Diagnostic Institute; normal range 14–50 pg mL–1). The intra-assay and interassay coefficients of variation for this assay are 5.8 and 9.9%, respectively.
Biochemical markers of bone turnover. Serum osteocalcin (OC) and serum total alkaline phosphatase (ALP) were measured to assess bone formation, whilst serum carboxy-terminal pyridinoline-cross-linked telopeptide of type I (ICTP) was measured as a marker of bone resorption.
ALP (normal range 50–300 UI L–1) was measured by an ‘optimized standard method’ (BM/Hitachi 917 Systems Pack, Boehringer Mannheim).
Measurement of OC (normal range [ng mL–1]: male, 1–12; premenopausal female, 1–8; postmenopausal female, 1–12) was carried out using a radioimmunoassay kit (CIS Diagnostici, Milano, Italy). The intra-assay and interassay coefficients of variation for this assay are 4.1 and 4.9%, respectively.
Serum ICTP was measured by a radioimmunoassay kit (ICTP, Orion Diagnostic, Finland; normal range 1.5–4 ng mL–1). The intra-assay and interassay coefficients of variation for this assay are < 6 and < 8%, respectively.
Bone mineral density values (g cm–2) and T scores were expressed as a means ± SD. The significance of the differences amongst CD, UC and control groups was calculated by one-way anova followed by Scheffe’s test. Continuous data which were normally distributed were presented as means ± SD and the statistical significance of differences between groups of patients and controls was assessed by using the Student’s unpaired t-test. Qualitative data were compared by the χ2 test. Relationships between two parameters were examined by calculating the Pearson’s correlation test. The dependence of lumbar spine and femoral neck BMD from different independent variables thought to be potential risk factors for bone loss, such as age, sex, BMI, disease duration and cumulative lifetime steroid dose, was assessed by a stepwise regression. Two-tailed significance tests were used in all statistical analyses. All statistical analyses were performed using Stat-View data analysis software (Abacus, Berkeley, CA).
From 1992 to 1995, 91 consecutive patients (46 male and 45 female) were included in the study. CD was diagnosed in 51 patients (21 male and 30 female, mean age 38.7 years), whilst UC was diagnosed in 40 patients (25 male and 15 female, mean age 34.4 years), with a disease duration of 83.5 ± 67.7 and 70.9 ± 44.0 months, respectively. No statistically significant difference was found between patients and controls with respect to age and sex distribution. Moreover, BMI was similar and there was no significant difference amongst the three groups in the number of postmenopausal women. Cumulative lifetime steroid dose was similar in CD and UC patients at the beginning of the study. Similarly, there was no significant difference between the CD and UC groups in the number of patients with active disease at the start of the study. In the CD group, 28 (54.9%) patients had ileocolonic disease, 16 (32%) had ileal disease only, and six (12%) had colonic disease only. Nineteen (37.2%) CD patients had had previous bowel resection, whilst no UC patient had had previous coloproctectomy.
Details of baseline characteristics of patients and controls are reported in Table 1.
Table 1. Baseline characteristics of patients and controls
BMI, body mass index; IC, ileocolonic disease; I, ileal disease; C, colonic disease; PC, total colitis; LSC, left-sided colitis.
Age (mean ± SD)
39.4 ± 11.6
38.7 ± 13.2
34.4 ± 12.5
BMI (kg m–2; mean ± SD)
23.1 ± 3.29
22.9 ± 3.13
23 ± 3.27
Duration of disease (months; mean ± SD)
83.5 ± 67.7
70.9 ± 44.0
Cumulative dose (g; mean ± SD)
6.160 ± 5.539
6.866 ± 5.846
< 10 mg day–1
> 10 mg day–1
Previous small bowel resection
Extent of disease
Table 2 reports the mean values ± SD of BMD, Z scores and T scores for lumbar spine and femoral neck, showing a reduced BMD in both CD and UC patients. However, when UC and CD were compared, no significant differences (P = 0.11) in BMD, Z scores and T scores were found, or within the patient subgroups according to disease location (P > 0.05). Lumbar spine T scores were significantly correlated with femoral neck T scores (r = 0.69, P = 0.001).
Table 2. Lumbar spine and femur bone mineral density (BMD) of Crohn’s disease (CD) and ulcerative colitis (UC) patients, and calculated Z scores and T scores (mean ± SD)
0.93 ± 0.203
–1.24 ± 1.65
–1.49 ± 1.46
0.899 ± 0.162
–1.44 ± 1.24
–1.67 ± 1.13
0.734 ± 0.141
–1.24 ± 1.38
–1.80 ± 1.36
0.754 ± 0.108
–1.20 ± 1.17
–1.60 ± 1.03
Figure 1 shows the distribution of femur T scores based upon the WHO guidelines. In the CD group, 4/51 (8%) had a normal T score, whilst 28/51 (55%) and 19/51 (37%) were osteopenic and osteoporotic, respectively. Correspondingly, 6/40 (15%), 27/40 (67%) and 7/40 (18%) UC patients were normal, osteopenic and osteoporotic, respectively.
No statistically significant difference in spine and femur T scores was found between postmenopausal women and other women (−1.44 ± 1.54 vs. −1.40 ± 1.35, P > 0.13). Similarly, there was no statistical difference in T scores in both CD and UC groups between patients with current disease activity and those in remission (CD: –1.39 ± 1.36 vs. –1.40 ± 1.33, P = 0.98; UC: –1.28 ± 1.21 vs. –1.23 ± 1.25, P = 0.91), and between patients who had undergone previous operations and others (–1.29 ± 1.10 vs. –1.21 ± 1.12, P = 0.84).
Stepwise regression showed that in the CD group, the femur T score was significantly inversely related to disease duration (r2 = 0.125, F = 6.06), whilst the spine T score was inversely related to age (r2 = 0.107, F = 5.49). In UC patients, spine T score was significantly related to sex, and was more negative in males (r2 = 0.3, F = 16.1), whilst femur T score was significantly related to sex (once again being more negative in male patients) and inversely related to the cumulative prednisone dose (r2 = 0.283, F = 7.3). Based upon the stepwise regression results, we evaluated differences in spine and femur BMD, Z and T scores between men and women in the CD and UC groups. Whilst in the CD group no statistically significant difference was found between men and women, amongst UC patients both spine and femur BMD, Z scores and T scores were significantly lower in men than in women ( Table 3).
Table 3. Lumbar spine and femoral neck bone mineral density (BMD), Z scores and T scores in male and female ulcerative colitis (UC) patients
Whilst calcium, PTH and 25(OH)D3 concentrations were not significantly different between patient groups and controls (P > 0.05), the 1,25(OH)D3 serum levels were significantly lower in both CD and UC patients when compared with controls (P < 0.05).
The biochemical markers of bone turnover were significantly different both between CD and UC patients and between the last two groups and controls. In particular, both markers of bone formation (ALP and OC) and bone resorption (ICTP) were significantly increased in UC patients, but not in CD patients and controls (P < 0.05).
Patients with IBD are at increased risk for low BMD with a prevalence of 30–75% in cross-sectional and prospective studies [1–11]. Bone loss is a major risk factor for osteoporotic fractures and, although the prevalence of fragility fractures associated with IBD is unknown, a number of reports have described clinical manifestations of severe osteoporosis in some patients, including young adults [1–3].
This cross-sectional study confirms many other reports of a high prevalence of low bone mineralization in IBD and shows that this feature is independent of site or type of disease (CD and UC). Despite the limited statistical power of the study to allow for subgroup analysis, the underlying mechanisms of this bone loss seem to be different in both CD and UC, as shown by the findings of an inverse correlation between BMD and disease duration and age in the CD group, and a negative correlation between the male sex and cumulative lifetime steroid dose in UC patients. Moreover, T scores were lower in men than in women in the UC group only. Finally, both markers of bone formation (OC and ALP) and bone resorption (ICTP), expressions of high bone turnover, are only significantly increased in UC patients, not in CD patients.
Two comparative studies on BMD in patients with CD and UC have been published. Ghosh et al.  reported that newly diagnosed untreated CD patients have decreased bone mineralization, whereas patients with newly diagnosed, untreated UC have normal BMD. Moreover, in this study, after steroid-induced remission, bone density did not result in further bone loss in a 1-year follow-up. Thus, the authors concluded that the disease itself and not corticosteroid therapy, is crucially determining low BMD in IBD. In a cross-sectional population-based study, Jahnsen et al.  confirmed this pattern in patients with long-standing CD and UC. They found that patients with CD had significantly reduced BMD compared with UC patients and healthy controls, and that there was no BMD difference between the last two groups. However, in contrast to Ghosh et al.’s study, use of corticosteroids was found to be associated with low BMD in CD but not in UC. The results of these studies suggest that disease diagnosis alone is the key factor in determining bone density in these patients. This hypothesis seems to be further supported by the results of a recent study in vitro, in which organ cultures of fetal rat parietal bones were treated with sera from subjects with CD, with UC and healthy controls. Compared with controls and UC, only serum from CD patients significantly decreased dry bone and calcium content during 96 h of culture. Moreover, histological evaluation of cultured bone showed marked differences between controls and CD patients, with the latter characterized by disorganization of mineral and osteoid and morphologically abnormal osteoblasts. Thus, these data suggest that circulating factors may affect osteoblasts and bone formation, leading to bone loss in CD but not in UC patients.
We aimed to evaluate possible pathogenetic differences in determining bone loss in IBD, comparing unselected patients with long-standing CD and UC, showing a reduced BMD in both patient groups. In our opinion, our results only apparently contrast with those observed by Ghosh et al. and Jahnsen et al., and with closer scrutiny our different conclusions may not be mutually exclusive. CD is a systemic disease with a long premorbid phase, whereas UC is a mucosal disease with an acute onset, often limited to distal colonic tracts. Moreover, CD also has important immunological differences from UC  and often, despite clinical remission, anatomical and biological activity may persist for a long period. The finding of an inverse relation between low BMD and disease duration and age in the CD group may signify that prolonged bone exposition to cytokines may be the primary contributory factor for low BMD and that other factors, although highly relevant, only additionally affect bone density in CD patients. On the contrary, the finding of an inverse correlation between reduced mineral content and cumulative lifetime steroid dose in UC may signify that at diagnosis UC patients may have a normal BMD, but that in time, with considerable corticosteroid use, the deleterious effect of steroids on bone becomes manifest. Thus, in Ghosh et al.’s study, the lack of evidence of low BMD in UC patients, after steroid treatment as well , may simply depend on the short duration of follow-up adopted.
We also observed an inverse correlation between BMD and the male sex, with T scores that were significantly lower in men than in women in UC but not in CD. Although we have not measured the serum levels of sex hormones, one can speculate that these findings may depend on the inhibitory effects of glucocorticoid on androgens . In men, testosterone deficiency may contribute to osteoporosis, and hypogonadism is an important and treatable cause of osteoporosis . In a very recent study in men with CD, 4/48 (8%) had low serum testosterone, all with long-standing ileocolonic disease requiring treatment with oral steroids. Although not statistically significant, testosterone was lower in the 12/48 (25%) men taking oral steroids at the time of study . These results emphasize the importance of including men as well as women with IBD in screening programmes for bone loss and in future pharmacological trials for the prevention and treatment of osteoporosis.
A further difference between CD and UC was the finding of a significant increase in serum levels of OC, ALP and ICTP in UC but not in CD patients. These results suggest that a high bone turnover in which the number of remodelling units is increased, resulting in an increase in the percentage of bone surface occupied by resorption cavities, may be the mechanism of bone loss, at least in UC patients. Several studies using biochemical markers of bone turnover have investigated the possible mechanism of bone loss in IBD, producing conflicting results. An increased bone resorption, as evidenced by elevated serum level of ICTP [4, 24] or the urinary deoxipyridinoline [6, 25], was found, without a consistent increase in any marker of bone formation. In contrast, a decreased serum level of OC below the normal range [24, 26 ] and a reduced bone formation on bone biopsy  were also detected. These conflicting results may reflect either differences in patient selection and methodology amongst the different studies or, more importantly, that more than the process contributes to bone loss in IBD. Furthermore, bone loss is likely to be phasic and the cumulative effect takes years, so that rates of loss demonstrated over one period of time cannot be assumed to reflect long-term changes. Thus, low bone formation and/or increased bone resorption could contribute to lower the BMD at different phases of the disease.
In our population, 1,25(OH)D3 serum levels were significantly lowered in both CD and UC patients compared with controls, despite normal levels of 25(OH)D3, calcium and PTH. The meaning of these results is unclear. Theoretically, if a reduced absorption of calcium occurs, during high inflammatory states, this should be associated with an increase in the levels of active vitamin D metabolite and PTH. On the other hand, animal studies suggest that steroids may impair vitamin D metabolism, and serum 1,25(OH)D3 levels are reduced in children and adolescents undergoing long-term treatment with steroids . This point needs to elucidated in further studies. However, it cannot be excluded that the difference between patients and controls may simply reflect seasonal variability, considering that patients and controls were not studied at same time.
In conclusion, our data show that long-standing, unselected IBD patients have a diffuse low bone density, the degree of which is not different in CD and in UC. However, the underlying pathogenetic mechanisms seem to be different: the disease itself appears to be the critical determinant of bone loss in CD, whilst long-term use of corticosteroids seems to be the major aetiological factor in UC. The finding of T scores that were more negative in male than in female UC patients suggests a potential inhibitory effect of corticosteroids on androgens.