SEARCH

SEARCH BY CITATION

Keywords:

  • bone mineral density;
  • Crohn's disease;
  • inflammatory bowel disease;
  • osteopenia;
  • osteoporosis

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Previous studies have confirmed that the prevalence of decreased bone mineral density is elevated in patients with inflammatory bowel disease. The objective of the current study was to determine the prevalence of osteopenia and osteoporosis in a cross-sectional outpatient population of 242 adult patients with Crohn's disease and to determine which clinical characteristics and serum and urine biochemical factors might be predictive of bone loss. Thirty-seven percent had normal bone density, 50.0% were osteopenic, and 12.9% were osteoporotic. Among the sites used to diagnose low bone mineral density, the femoral neck demonstrated the highest prevalence of osteopenia and the ultra-distal radius the highest prevalence of osteoporosis. However, low bone mineral density at one site was always predictive of low bone mineral density at the other. Corticosteroid use during the year before assessment was found to be consistently predictive of low bone mineral density in males but not in females. In contrast, low body mass index and high platelet counts were consistently predictive of low bone mineral density in females but not in males. Disease location, smoking, and age were not predictive of changes in bone mineral density.

Several studies within the past decade have revealed that the incidence of low bone mineral density (BMD) is increased in patients with inflammatory bowel disease, with a higher prevalence in Crohn's disease (CD) than in ulcerative colitis (UC). 1–5 The mechanism underlying the lower BMD has not been clearly understood, but a number of factors have been implicated. These include disease activity, corticosteroid therapy, calcium and vitamin D deficiency, acute inflammatory cytokine action on osteoclast and osteoblast activity, sex hormone deficiency, smoking, and overall poor nutrition. 2,3,6–14 Study results have not been consistent, however, and it is likely that the cause is multifactorial.

Acute intestinal inflammation is postulated to affect bone mineral density in CD patients as a consequence of an increase in circulating proinflammatory cytokines. 10,15 These cytokines have been demonstrated to have an effect on bone metabolism in vitro, inhibiting osteoblast function and encouraging bone resorption. 10 High serum levels of IL-6, for example, have been found in osteoporotic CD patients, compared with non-osteoporotic patients. 16 In addition, acute intestinal inflammatory events can induce central hypogonadism in both males and females and can cause amenorrhea in females. 11,17 Amenorrheal states have been strongly associated with osteoporosis, and hormone replacement therapy is effective for the treatment of bone loss in postmenopausal women with CD. 18 To date, few studies have been attempted to determine whether an association between markers of inflammation and bone loss in Crohn's disease exists. Knowledge of such an association would be helpful in determining which patients with Crohn's disease are at risk for developing osteopenia and osteoporosis.

Osteopenia is defined by the World Health Organization (WHO) as a T-score between −1 and −2.5, and osteoporosis is a T-score less than −2.5, the T-score being the number of standard deviations (SD) of the patients BMD from the mean peak value for a reference population of the same sex and race. 19,20 Additionally, a low BMD measurement at an individual site will generally reflect the fracture rates at the respective site. The femoral neck and total hip measures have been identified as indicators of hip fracture, whereas the lumbar spine BMD is an indicator of vertebral fracture risk. 21,22 Similarly, the ultra-distal radius BMD indicates relative risk of Colle's fracture. 23 Indeed, the studies previously cited have reported that between 36% and 55% of CD patients are osteopenic, whereas between 6% and 58% are osteoporotic. 6,16,24–26 This high prevalence of osteopenia and osteoporosis in CD patients is likely coupled with an increased risk of fracture 2,27–29 and necessitates the development of a clinically effective approach to determining which patients with CD are at highest risk of bone loss.

To meet this objective, we must first determine what patient characteristics are associated with decreased BMD in CD patients. Characteristics such as low body mass index (BMI), age, gender, high levels of serum alkaline phosphatase and albumin, high levels of urine N-telopeptide cross-linked type I collagen (N-telopeptide), low serum magnesium, corticosteroid use, and smoking have previously been reported to correspond with low BMD. 1–4,6,7,13,14,26,30–34 Unfortunately, many of these studies presented contradictory results, likely as a consequence of small sample sizes and heterogeneous disease populations (Crohn's disease and ulcerative colitis).

The objective of the current study, thus, is to determine the prevalence of osteopenia and osteoporosis in a large cross-sectional population of adult patients with Crohn's disease and to assess what patient characteristics and serum and urine biochemical factors might be used to predict bone loss.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Patients

Between 1998 and 2000, 224 patients with Crohn's disease who had attended the Inflammatory Bowel Disease Clinic at the University of Alberta Hospital (Edmonton, Canada) were consecutively enrolled in this study. An additional 18 patients from Mount Sinai Hospital (Toronto, Canada) were also enrolled, giving a total of 242 patients. Informed consent from each patient was received in writing. Crohn's disease was diagnosed on the basis of endoscopic, radiologic, and histologic examination.

The following exclusion criteria were applied at baseline: Age < 18 years, known bone disorders other than osteoporosis (hyperparathyroidism, Paget's disease, renal osteodystrophy, and documented osteomalacia), abnormal thyroid function, significant renal impairment (serum creatinine ≥ 2× normal), clinical short bowel syndrome, patients on total parenteral or enteral nutrition, and spinal anatomy that would not allow adequate assessment of lateral spine x-rays using dual-energy x-ray absorptiometry (DXA). In addition, patients who had received (1) bisphosphonate in the 24 months prior to the data collection program entry, (2) fluoride supplement in the 24 months prior to entry, (3) calcium supplements of more than 1.0 g/day, or (4) vitamin D supplements greater than 1000 IU/day in the 6 months prior to entry were excluded from the study. Postmenopausal females were not excluded from the study, and females on oral contraceptive use or hormone replacement therapy were not excluded as long as the therapy was implemented 3 months prior to baseline and was continued throughout the study period.

A baseline questionnaire administered by the study nurse documented patient demographics and disease information, age at baseline, diagnosis date, gender, smoking status, disease activity in the past year, and corticosteroid use in the past year. Other baseline information was gathered through review of the patients' records, blood and urine analysis, and DXA analysis. Weight and height were measured and BMI (in kg/m2) subsequently determined. Disease site was recorded as colonic, ileal, ileocolonic, or jejunal/duodenal. Patients were then separated into three groups: those with normal bone densities (T ≥−1.0), those with osteopenia (−1.0 < T ≤−2.5), and those with osteoporosis (T < −2.5). 19

Bone Mineral Density Measurements

The bone mineral densities (g/cm2) of the lumbar spine (L1-L4), the femoral neck, total hip, and ultra-distal radius, were measured at baseline by DXA (Hologic 4500, Waltham, MA, USA) using standard protocols. Vertebral BMD values were normalized using the Hologic reference database; hip BMD values were normalized using the National Health And Nutrition Examination Study III (NHANES III) database. 3 The coefficients of variation for the lumbar spine, femoral neck, total hip, and ultra-distal radius were 1.2%, 0.8%, 1.0%, and 1.0%, respectively. Osteopenia (−1.0 < T ≤ −2.5) and osteoporosis (T < −2.5) were diagnosed using the lowest T-score of the following: (1) the lumbar vertebrae (L1-L4 inclusive), (2) total hip, or (3) femoral neck. Total body height was measured using a wall-mounted digital stadiometer (Holtain Ltd., Crymych, Dyfed, UK).

Biochemical Measurements

Blood samples were taken at baseline to determine the following serum values: serum alkaline phosphatase (IU/L), phosphorous (mmol/L), calcium (mmol/L), parathyroid hormone (pmol/L), 25-OH vitamin D (nmol/L), platelet count (×109 /L), white blood cell (WBC) count (×109 /L), hemoglobin (g/L), C-reactive protein (CRP) (mg/L), ferritin (μg/L), carotene (μmol/L), vitamin B12 (pmol/L), RBC folate (nmol/L), testosterone (nmol/L), follicle stimulating hormone (FSH) (U/L), luteinizing hormone (LH) (U/L), estradiol (pmol/L), and thyroid stimulating hormone (TSH) (U/L). Urine samples were also collected for a 24-hour period at baseline to measure urine N-telopeptide levels (nmol/mmol urine creatinine). CRP, WBC count, and platelet count served as markers of inflammation and Crohn's disease activity. 36,37 Urinary N-telopeptide was measured as a marker of bone resorption. 33,38

Statistical Analysis

Descriptive statistics were reported as a mean ± SD (SD) for continuous variables that were normally distributed, unless otherwise stated. For comparison of T-scores between two groups, a Student's t test was performed. This test was performed for males versus females, corticosteroid versus non-corticosteroid use in the past year, and disease activity (≤2 flares vs >2 flares). If the data was not normally distributed, then the Mann-Whitney rank test was employed. T-scores were then corrected for covariates and significance of differences in T-score between groups reassessed, to verify results. Comparisons of the clinical and biochemical characteristics between the normal BMD, osteopenic, and osteoporotic groups was performed using a one-way analysis of variance (ANOVA), followed by the Kruskal-Wallis one-way ANOVA on ranks if the data were not normally distributed. Differences in T-score between different disease locations were assessed using one-way ANOVA. Correlations between clinical variables and bone mineral density in the male and female patient groups were performed using Pearson's coefficient of correlation. To determine if any variables were predictive of BMD, a stepwise regression analysis was performed, setting clinical factors as independent variables and each skeletal BMD measurement site as the dependent variable.

For all tests, significance was defined as p < 0.05 (95% confidence interval), with normality and variance defined at p < 0.01. The statistical program SPSS (Statistical Package for the Social Sciences, ver. 11.0, 2001) was used for all statistical procedures.

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Patient Clinical Characteristics

Baseline patient clinical characteristics of the study patients are given in Table 1. Forty-seven (20.8%) patients had colonic CD, 87 (38.5%) patients had ileal CD, and 90 (39.9%) patients had ileocolonic CD. Only two patients displayed jejunal/duodenal CD. Eighty-three (36.9%) patients reported smoking, and 122 (51.7%) reported using corticosteroids in the past year. One hundred eighty patients (80.4%) reported 0 to 2 flares of disease activity requiring a physician's visit in the past year, whereas 42 (18.8%) reported between 3 and 6 flares, and 2 patients reported between 7 and 10 flares in the past year.

Table 1. Baseline Clinical Characteristics
Clinical Characteristic n = 242 
  1. Values are reported as mean ± standard deviation, unless otherwise stated.

Age (years)38.7 (±12.3)
Gender (M/F)108/134
Age at diagnosis (years)27.6 (±12.0)
Years since diagnosis11.1 (±8.7)
Disease location(% of patients)
 Colon 20.8
 Ileum 38.5
 Colon & ileum 39.9
 Jejunum/duodenum 0.8
% Smokers (at any time)36.9
BMI (kg/m2)25.0 (±4.6)
No. of flares requiring a physician's visit in last year(% of patients)
 ≤2 80.4
 3–6 18.8
 7–10 0.8
% of patients who used steroids in last year51.7
Bone Mineral Density n/(%) 
 Normal 89/(37.1)
 Osteopenia120/(50.0)
 Osteoporosis 31/(12.9)

Bone Mineral Density Characteristics

Eighty-nine patients (37.1%) had normal bone density, 120 patients (50%) were osteopenic, and 31 patients (12.9%) were osteoporotic (Table 1). Osteopenia was most pronounced at the femoral neck (46.0%), followed by the ultra-distal radius (44.8%), lumbar spine (38.3%), and the total hip (27.8%). Osteoporosis was most pronounced at the ultra-distal radius (9.5%), followed by the lumbar spine (7.9%), femoral neck (5.9%), and the total hip (3.0%).

Clinical characteristics of the normal BMD, osteopenic, and osteoporotic patient groups are given in Table 2. There is a higher proportion of females in the normal density group (64%) than in the osteopenic (47.5%) and osteoporotic groups (58.1%) (difference = 16.5% and 5.9%, respectively; p = 0.037 and 0.049, respectively). There were no significant differences in either the mean age or disease location among the three groups (data not shown). Corticosteroid use was higher in each of the osteopenic (50.8%) and osteoporotic groups (66.7%), as compared with the normal group (44.2%) (p = 0.021 and p < 0.001, respectively), and BMI was significantly lower in the osteopenic (24.5 ± 4.2 kg/m2) and osteoporotic groups (23.3 ± 4.8 kg/m2), as compared with the normal group (26.5 ± 4.8 kg/m2) (p = 0.004 and 0.002, respectively).

Table 2. Comparison of Baseline Clinical Characteristics and Bone Mineral Density
 Bone Mineral Density
Clinical CharacteristicNormalOsteopenicOsteoporotic
  • Values are reported as mean ± standard deviation, unless otherwise stated. Abbreviations: BMI, body mass index. WBC, white blood cell, LH, luteinizing hormone.

  • *

    *Denotes statistical significance relative to normal BMD group (p< 0.05, Cl = 95%).

  • †Denotes statistical significance relative to osteopenic group (p <; 0.05, Cl =95%).

n8912031
Gender (M/F) (%)36.0/64.052.5/47.5*41.9/58.1*
Smokers (%)43.8%33.0%34.5%
Corticosteroid use in last year (%)44.250.8*66.7*
Age (year)36.5 ± 10.739.5 ± 12.342.2 ± 15.7
Disease duration (year)10.2 ± 7.911.6 ± 9.311.7 ± 8.6
BMI (kg/m2)26.5 ±4.824.5 ±4.2*23.3 ±4.2*
Platelets (x109*/L)256.4 ± 73.0281.4 ± 79.3309.0 ± 91.7+
WBC(x109/L)7.3 ± 2.37.3 ± 2.57.9 ± 2.9
LH (U/L)7.5 ± 9.318.5 ± 21.521.5 ± 20.3
N-telopeptide/urine creatinine (nmol/mmol)42.8 ± 24.044.2 ± 27.172.8 ±90.4 *
T-score   
 L1-L41.08 ± 0.08–1.16 ± 0.71–2.45 ± 0.91
 Femoral neck0.89 ± 0.09–1.30 ± 0.56–2,39 ± 0.55
 Total hip1.05 ± 0.09–0.78 ± 0.53–1.89 ± 0.73
 Ultra-distal radius–0.42 ±0.85–1.43 ±0.83–2.24 ± 0.84

When biochemical parameters were compared, the platelet count was significantly higher in the osteoporotic group than in the group with normal bone density (309 ± 91.7 vs 256.4 ± 73 × 109 /L, respectively, p = 0.004), as was urine N-telopeptide (72.8 ±90.4 vs 42.8 ± 24.0 nmol/mmol urine creatinine respectively, p = 0.002).

No other clinical or biochemical parameter significantly differed among the low bone density groups, as compared with the normal BMD group.

Effect of Corticosteroid Use and Gender on Bone Mineral Density

The mean T scores for the lumbar spine (L1-L4), femoral neck, total hip and ultra-distal radius are given in Table 3 for corticosteroid versus noncorticosteroid use in the year preceding baseline and for males versus females.

Table 3. Corticosteroid Use During the Preceding Year and Gender Characteristics Relative To Bone Mineral Density
 Bone Mineral Density T-Score
CharacteristicLl-L4Femoral NeckTotal HipUltra-Distal Radius
  • Values are mean ± standard deviation.

  • *

    *Denotes corticosteroid use or no corticosteroid use in the preceding year.

  • †Denotes statistically relevant difference compared to no steroid use (p < 0.05, CI = 95%).

  • ‡Denotes statistically relevant difference compared to male non-steroid use (p < 0.05, CI = 95%).

  • §

    §Denotes statistically relevant difference compared to females (p < 0.05, Cl = 95%).

Corticosteroid use*    
 All(n= 123)−1.07 1.12−1.06 ±1.07−0.67 ± 1.01*−1.29 ± 1.07
 Males (n = 55)−1.46 ±1.05−1.24 ±0.88−0.77 ±0.84−1.53 ± 1.05
 Females (n = 68)−0.74 ±1.08−0.91 ±1.19−0.59 ± 1.14−1.10 ± 1.06
No corticosteroid use*    
 A11(B= 116)−0.59 ± 1.16−0.73 ± 1.09−0.20 ± 1.05−0.95 ± 0.99
 Males (n = 51)−0.79 ± 1.09−0.73 ± 0.85−0.17 ± 0.78−1.07 ± 0.92
 Females (n = 65)−0.44 ± 1.19−0.73 ± 1.25−0.21 ± 1.22−0.86 ± 1.04
Males (n = 106)−1.13 ± 1.11§−1.00 ± 0.89−0.49 ± 0.85−1.28±1.01§
Females (n= 133)−0.59 ± 1.14−0.81 ± 1.23−0.38 ± 1.22−0.98 ± 1.05

When both male and female patients were combined, significant differences in BMD T scores at all sites were found in those patients who had used corticosteroids in the preceding year relative to those who had not (p = 0.001, 0.021, <0.001, and 0.016, for L1-L4, femoral neck, total hip, and ultra-distal radius, respectively). Male corticosteroid users showed significantly lower BMD T-scores than male noncorticosteroid users (p = 0.002, 0.003, <0.001, and 0.024 for L1-L4, femoral neck, total hip, ultra-distal radius, respectively). Female corticosteroid users did not have a statistically lower BMD T-score at any skeletal site when compared with female noncorticosteroid users. Interestingly, female noncorticosteroid users were, on average, older than the female corticosteroid use group (42.8 vs 34.0 years, p < 0.001)

When T-scores for the entire population were adjusted for age, BMI, platelet count, urine N-telopeptide, and corticosteroid use continued to exhibit statistically significant lower mean T-scores at the lumbar spine (−1.008 vs −0.614, respectively, p = 0.017), femoral neck (−1.091 vs −0.719, respectively, p = 0.011), total hip (−0.609 vs −0.198, respectively, p = 0.003), and ultra-distal radius (−1.274 vs −0.961, respectively, p = 0.033). Male corticosteroid users exhibited statistically significant lower mean T-scores when compared with male noncorticosteroid users at the lumbar spine (−1.316 vs −0.735, respectively, p = 0.022), femoral neck (−1.122 vs −0.716, respectively, p = 0.037), total hip (−0.606 vs −0.176, respectively, p = 0.018), but not the ultra-distal radius (−1.376 vs −1.080, respectively, p = 0.192). Also, males continued to exhibit significantly lower mean lumbar T-scores than females (−1.055 vs −0.568, respectively, p = 0.003), though not at the ultra-distal radius (−1.242 vs −0.993, respectively, p = 0.090).

Male corticosteroid and male noncorticosteroid users were compared for clinical and biochemical characteristics to identify possible causative factors that might account for the difference in BMD between the two groups. The statistically significant clinical and biochemical characteristics for the two groups are given in Table 4. Male corticosteroid users exhibited lower hemoglobin concentrations and higher WBC and platelet counts (p = 0.01, 0.001, and 0.013, respectively).

Table 4. Clinical and Biochemical Characteristics Between Males Who Had Used Corticosteroids in the Preceding Year and Those Who Had Not
CharacteristicMale Corticosteroid Use Group (n=55)Male Noncorticosteroid Use Group (n=51)
  • Values are expressed as means ± standard deviation. Abbreviations: WBC. white blood cell.

  • *

    *Denotes statistical significance relative to male non-steroid use group (p < 0.05, CI = 95%).

Hemoglobin (g/L)141.2 ±14.14 *148.0 ±11.2
Platelet count (x109/L)279.2 ± 72.3*243.6 ±69.3
WBC count (x109L)8.2 ±2.7*6.6 ± 2.0

Correlations and Predictive Factors of Low BMD

Correlation coefficients between clinical characteristics and BMD in the male (Table 5) and female (Table 6) CD populations were determined to identify predictive factors of low BMD. Most strikingly for males, corticosteroid use was significantly negatively correlated with each of the skeletal BMD measurement sites (r = −0.301, −0.286, −0.348, and −0.233 for the lumbar spine, femoral neck, total hip, and ultra-distal radius T-scores, respectively, p < 0.05). In contrast, most striking for females, BMI positively correlated with each skeletal BMD measurement site (r = 0.394, 0.351, 0.406, and 0.339 for lumbar, femoral neck, total hip, and ultra-distal radius T-scores, respectively, p < 0.05).

Table 5. Pearson Correlation Coefficients in 108 Males, Comparing Age, Body Mass Index (BMI), Corticosteroid Use in the Preceding Year, Platelet Count, White Blood Cell Count (WBC), and Urine N-Telopeptide, to Baseline T-Scores of the Lumbar Spine, Femoral Neck, Total Hip, and Ultra-Distal Radius
CharacteristicLumbar L1-L4 T-ScoreFemoral Neck T-ScoreTotal Hip T-ScoreUltra-Distal Radius T-Score
  • *

    *Denotes p<0.05.

Age (years)0.116−0.295*−0.0340.032
BM1 (kg/m2)0.1680.0820.245*0.086
Corticosteroid use−0.301*−0.286*−0.348*−0.233*
WBC count (x109)0.199*−0.092−0.0340.009
Platelet count (x109)−0.120−0.088−0.177−0.024
N-telopeptide (nmol/mmol urine creatinine)−0.256*−0.115−0.291*−0.302*
Table 6. Pearson Correlation Coefficients in 134 Females, Comparing Age, Body Mass Index (BMI), Corticosteroid Use in the Preceding Year, Serum Albumin, Platelet Count, Hemoglobin, and Urine N-Telopeptide, to Baseline T-Scores of the Lumbar Spine, Femoral Neck, Total Hip, and Ultra-Distal Radius
CharacteristicLumbar L1-L4 T-ScoreFemoral Neck T-ScoreTotal Hip T-ScoreUltra-Distal Radius T-Score
  • *

    *Denotes p <0.05.

Age (years)−0.019−0.262*−0.0850.032
BMI (kg/m2)0.394*0.351*0.406*0.339*
Corticosteroid use−0.138−0.073−0.158−0.117*
Albumin (g/L)0.1050.1780.1680.220*
Hemoglobin (<g/L)0.219*0.1670.200*0.207*
Platelet count (x109)−0.299*−0.263*−0.287*−0.336*
N-telopeptide (nmol/mmol urine creatinine)−0.270*−0.143−0.206*−0.407*

In males, N-telopeptide correlated negatively with lumbar spine (r = −0.256, p < 0.05), total hip (r = −0.291, p < 0.05), and ultra-distal radius (r = −0.302, p < 0.05) T-scores. Similarly, in females, N-telopeptide correlated negatively with lumbar spine (r = −0.270, p < 0.05), total hip (r = −0.206, p < 0.05), and ultradistal radius (r = −0.407, p < 0.05) T-scores. N-telopeptide did not demonstrate statistical significance when correlated with femoral neck T-score in males and females (r = −0.115, p = 0.1, r = −0.143, p = 0.12, respectively). In females, but not in males, platelet counts were negatively associated with BMD (r = −0.299, −0.263, −.287, and −0.336, for the lumbar, femoral neck, total hip, and ultra-distal radius T-scores, respectively, p < 0.05).

When forward stepwise regression analysis was performed in the male CD group, corticosteroid use and serum vitamin B12 were found to be predictive of lumbar spine T-score (p = 0.005 and 0.011, respectively, r = 0.408). Femoral neck T-score was predicted by corticosteroid use, age, and BMI (p = <0.001, p = 0.009 and 0.017, respectively, r = 0.475). Total hip T-score was predicted by corticosteroid use and urine N-telopeptide (p = 0.003 and 0.02, respectively, r = 0.420). Ultra-distal radius T-score was predicted only by urine N-telopeptide (p = 0.005, r = 0.302).

In the female CD group, forward stepwise regression demonstrated that Lumbar spine T-score was predicted by BMI and platelet count (p = <0.001 and p = 0.001, respectively, r = 0.480). Femoral neck T-score was predicted by age, BMI, and platelet count (P < 0.001 for all, r=0.570). Total hip T-score was predicted by age, BMI, platelet count, and corticosteroid use (p < 0.001, p < 0.001, p = 0.005 and 0.026, respectively, r = 0.592). Ultra-distal radius T-score was predicted by BMI, platelet count, and urine N-telopeptide (p = <0.001, p = 0.042 and 0.001, respectively, r = 0.549).

To determine if measurement of BMD at one skeletal site could predict BMD measured at another, measurements of BMD at the different sites were compared by regression analysis. Lumbar spine T-score was predicted by total hip T score and ultra-distal radius T score (p < 0.001, p = 0.021, respectively, r = 0.857). The femoral neck T-score could be predicted by total hip T-score and UD radius T-score (both p < 0.001, r = 0.767). Total hip T-score was predicted by lumbar spine T-score and femoral neck T-score (both p < 0.001, r = 0.776). Finally, ultra-distal T-score was predicted by total hip T-score and lumbar spine T-score (both p < 0.001, r = 0.690).

No statistically significant differences in BMD were found when smokers were compared with nonsmokers or when the number of disease flares in the past year was assessed. Twenty-nine females in our study were determined to be postmenopausal. Of these, 12 were on continuous hormone replacement therapy. The prevalence of osteopenia and osteoporosis in the females using hormone replacement therapy did not differ significantly from that of the rest of the female study population (51.7% and 20.7% vs 42% and 12% osteopenia and osteoporosis, respectively, p = 0.13).

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

This study, the largest of its kind, confirms the previous reports of the prevalence of osteopenia and osteoporosis in patients with Crohn's disease. Our findings of 50% osteopenia and 12.9% osteoporosis within the patient population lie in the ranges of 36.4% to 55% osteopenia and 5.5% to 57.6% osteoporosis seen in previously published results. 6,16,24–26 Several reasons account for the large ranges in osteopenia and osteoporosis reported in these previous studies. Small sample populations, the inclusion of both UC and CD patients into a single analysis, the inclusion of patients with ileal resections, the use of different definitions of osteoporosis and osteopenia, the use of Z-scores instead of T-scores, and the use of varying measurement techniques all contribute to these differences.

We found osteopenia to be more prevalent in the femoral neck (46%) than in the lumbar spine (38.3%), while in contrast, osteoporosis was more common in the lumbar spine (7.9%) than in the femoral neck (5.9%). This is in agreement with Bjarnson et al, who also reported a higher prevalence of hip osteopenia (49% hip vs 36% vertebral osteopenia). 26 The same study, however, reported a much higher incidence of osteoporosis (29% hip and 18% vertebral).

Our study demonstrated that corticosteroid use in the preceding year was associated with a significant decrease in BMD at all sites measured. Interestingly, this was true only for the male patient population. Females did not exhibit differences in BMD when corticosteroid use was compared. However, the female corticosteroid use group was younger than the group that had not used corticosteroids, and this difference may have influenced the outcome. Two previous studies, one assessing BMD of the spine and femoral neck of 120 Caucasian patients in the United Kingdom and the other assessing BMD of 60 patients with CD and 60 with ulcerative colitis, also observed BMD loss in the male corticosteroid use population. 4,39 Conversely, other studies have found that being female was associated with an increased the risk of low BMD in the case of CD patients. 7,31,34 One of these female-dominant studies, however, compared the patient population to healthy controls, rather than to other female CD patients not taking corticosteroids 31; whereas the other two reported results from a small sample patient population (26 and 55 patients, respectively). 7,34 Though the current study suggests that corticosteroid use has a more deleterious effect on bone mineral density on males than on females, the reason for this is not clear. One possibility involves the fact that corticosteroid therapy can suppress gonadotropin and testosterone levels, thus interfering with the inhibitory effects of testosterone on IL-6, allowing IL-6 to stimulate osteoclast activity. 11,40 However, in this study, mean testosterone levels in the male corticosteroid group did not differ significantly from those in the male non-corticosteroid group (data not shown).

Another possible explanation for the lower BMD in the male corticosteroid using population involves the fact that corticosteroid use reflects disease activity and severity. Indeed, we found that, compared with male noncorticosteroid users, males on corticosteroid therapy exhibited higher WBC and platelet counts, and lower serum hemoglobin levels, all of which indicate active and severe disease. Disease activity is believed to lead to bone loss in CD patients by increasing circulating levels of IL-1, IL-6, and TNF-α, thus enhancing osteoclast activity and inhibiting osteoblast activity. 10,40

Taken together, our results clearly demonstrate decreased bone mineral density in patients with Crohn's disease. Several studies in non–Crohn's disease patients has identified that reduced BMD is highly predictive of fracture risk, in both men and women 28,41–44. However, the relationship between low bone mineral density and fracture rate in patients with Crohn's disease remains controversial. In a Danish cohort of Crohn's disease patients fracture risk, as determined by ICD coding, was slightly elevated with an incidence rate ratio of ˜1.15. 45 Similarly, a Canadian population-based cohort study demonstrated an incidence rate ration of ˜1.74 and 1.59 for the spine and hip, respectively. 27 In contrast, in a population-based inception cohort of patients with Crohn's disease, the risk of fracture was not elevated relative to controls. 46 The current study did not assess fracture rate and, thus, cannot comment on the ability of low BMD to predict fracture risk in Crohn's disease patients.

Regression analysis to determine predictors of low BMD identified corticosteroid use as the factor that most strongly predicted low BMD at each skeletal site in the male patient population. The predictive nature of corticosteroids use on BMD, in this study, was identified by determination of corticosteroid use in the past year and did not require determination of cumulative corticosteroid doses. Other weaker predictors of low BMD in males included BMI, older age, vitamin B12, and higher N-telopeptide excretion. In females, the most significant predictors of low BMD were BMI, platelet count, older age, corticosteroid use (only at the total hip), and N-telopeptide (only at the ultra-distal radius). There have been similar findings of an association of corticosteroid use, BMI, urine N-telopeptides, and age to low BMD 1–4,6–14,24–26,30,33,41; however, this study was the first to demonstrate that platelet count could predict low BMD in the female population. Platelets are involved in the pathogenesis of CD and contribute to it by releasing inflammatory mediators. 42 Furthermore, active CD is associated with thrombocytosis, and thus platelet count serves as a marker of disease activity. 37,38,47

The finding that urine N-telopeptide levels correlated with the lumbar spine, total hip, and ultra-distal radius T-scores of both male and female CD patients suggested that the mechanism for bone loss in patients with Crohn's disease involved increased bone resorption. Other studies have identified a strong relationship between markers of bone resorption and bone loss in patients with Crohn's disease. 33,48,49 Although high N-telopeptide levels predicted low BMD at the total hip and ultra-distal radius in males, and at the ultra-distal radius in females in our study, it failed to do so at the lumbar spine and femoral neck in both males and females with CD. This suggests that increased bone resorption is only one of many mechanisms involved in the pathogenesis of osteoporosis in Crohn's disease.

The current study did not exclude postmenopausal women. It is well-known that postmenopausal women suffer greater bone loss because of decreased estrogen production. In our study, 21.6% of the female patients were postmenopausal. However, this group's bone density distribution did not differ appreciably from that of the entire baseline group and thus did not likely influence the BMD results of the female population. Furthermore, we excluded serum LH, FSH, and estradiol from the correlation and regression analyses. It was our assertion that these variables were not linearly related to BMD, but rather increased in a stepwise fashion, according to age and menopausal status of our female patient group.

Finally, it is difficult to dissect the influence of disease activity versus corticosteroid use as the primary the causative factor for bone loss in CD patients. The reason for this is that patients with active CD are most likely to be on corticosteroid therapy, and vice versa. A longitudinal analysis assessing CD patients from the time of diagnosis, and possessing noncorticosteroid therapies for active inflammatory disease, is essential in determining the exact role of these factors in Crohn's disease-associated bone loss. Nevertheless, the results of our study are significant in that we were able to assess several clinical and biochemical factors simultaneously, with a large patient population. In addition, we separately analyzed the male and female patient populations, taking into consideration that inherent differences between the two groups may exist and can adversely affect our results if analyzed as a whole. This may explain the conflicting results reported by previous studies, as they failed to separate the patient population into female and male groups for analysis.

In conclusion, this study focused on identifying risk factors for bone loss in a cross-sectional survey of 242 patients with Crohn's disease. Clinical and biochemical characteristics were assessed to determine associations with bone mineral density at four skeletal sites. Thirty-seven percent had normal bone density, 50.0% were osteopenic, and 12.9% were osteoporotic. Among the sites used to diagnose low bone mineral density, the femoral neck demonstrated the highest prevalence of osteopenia and the ultra-distal radius the highest prevalence of osteoporsis. However, low bone mineral density at one site was always predictive of low bone mineral density at the other. Corticosteroid use during the year before assessment was found to be consistently predictive of low bone mineral density in males but not in females. In contrast, low body mass index and high platelet counts were consistently predictive low bone mineral density in females, but not in males. Disease location, smoking, and age were not predictive of changes in bone mineral density.

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References