Predictors of Osteoporosis and Vertebral Fractures in Patients Presenting with Moderate-to-Severe Chronic Obstructive Lung Disease

Abstract

Bone mineral density (BMD) alone does not reliably predict osteoporotic fractures. The Fracture Risk Assessment Tool (FRAX) was developed to estimate the risk of fracture in the general population. This study was designed to identify predictors of osteoporosis and vertebral fractures in patients presenting with chronic obstructive pulmonary disease (COPD). We studied 85 patients (mean age = 75 years; 92% men) with moderate to very severe COPD. Osteoporosis and vertebral fractures were diagnosed with dual energy X-ray absorptiometric scan and vertebral X-rays, respectively. Patient characteristics, including age, gender, body mass index (BMI), and results of pulmonary function tests, chest computed tomography scan, blood and urinary biomarkers of bone turnover were recorded, and a FRAX score was calculated by a computer-based algorithm. Osteoporosis, defined as a T score < –2.5, found in 20 patients (24%), was associated with female gender, BMI, dyspnea scale, long-term oxygen therapy (LTOT), vital capacity (VC), emphysema score on computed tomography, measurements of serum and urinary biomarkers of bone turnover. Vertebral fractures, diagnosed in 29 patients (35%), were strongly correlated with age, LTOT, VC, and forced expiratory volume in 1 sec, treatment with oral corticosteroid or warfarin, and weakly associated with the presence of osteoporosis. There was no correlation between FRAX score and prevalence of vertebral fractures, suggesting that neither BMD alone nor FRAX score would predict the presence of vertebral fractures in COPD patients. A disease-specific algorithm to predict osteoporotic fractures is needed to improve the management of patients suffering from COPD.

Keywords :

• Bone mineral density
• Fracture Risk Assessment Tool
• Long-term oxygen therapy
• Chronic respiratory failure
• Corticosteroid

Introduction

Osteoporosis, a disorder in which bone mineral density (BMD) is decreased and the risk of fracture is increased, is a major co-morbidity in patients suffering from chronic obstructive lung disease (COPD). Osteoporotic fractures of the spine may be particularly detrimental in these patients, as deformation of the thorax and secondary kyphosis may further deteriorate their lung function. It is estimated that, in osteoporotic women, each vertebral fracture decreases the forced vital capacity (FVC) by nearly 10% (1), causing significant decreases in both FVC and forced expiratory volume in 1 sec (FEV₁) compared with healthy women or women presenting with uncomplicated osteoporosis (2). Therefore, the prevention of osteoporotic bone fractures is an important component of the management of patients presenting with COPD.

Although an early diagnosis and treatment of osteoporosis lowers the risk of vertebral fractures, they do not guarantee complete protection, as fractures may occur in presence of a normal BMD (3–6). Therefore, a Fracture Risk Assessment Tool (FRAX) was developed by the World Health Organization (WHO) to estimate the risk of fracture.(7), FRAX calculates the 10-year risk of fracture, using up to 10 variables, including age, gender, body mass index (BMI), personal history of fracture, history of parental hip fracture, current smoking, consumption of >60 g/day of alcohol, treatment with corticosteroids, history of rheumatoid arthritis or other disorders strongly associated with osteoporosis, and BMD, if available (7).

It is, however, not clear whether this tool, originally developed for post-menopausal women, reliably predicts the risk of fractures in patients suffering from COPD. Considering a) the detrimental effects of COPD on physical activity and nutritional status, and b) the pharmaceuticals used for its treatment, such as inhaled or oral corticosteroids, we expected to identify disease-specific risk factors for osteoporotic bone fractures in this population. This study a) examined the predictors of vertebral fractures in patients with moderate to very severe COPD, and b) compared these factors with the predictors of a decreased BMD.

Methods

Study population

We recruited 85 patients who presented to the pulmonary clinic of Keio University Hospital for management of moderate to very severe COPD, diagnosed by 1) a post-bronchodilator FEV₁/FVC <0.70, 2) a ≥10 pack-years smoking history, 3) age >40 years, and 4) a post-bronchodilator FEV₁ <70% of predicted or under long-term oxygen treatment (LTOT). The severity of COPD was defined according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD, 2009) classification (8). Patients treated with bisphosphonates, vitamin D, or vitamin K were excluded.

Personal histories of rheumatoid arthritis, diabetes mellitus, bone fractures, current or previous treatment with corticosteroids or warfarin potassium, smoking and alcohol habits, modified Medical Research Council (MMRC) grade 0–4 dyspnea scale, and family history of femoral neck fracture, were ascertained by a questionnaire. Oral corticosteroid use was defined as > 3 months use of predonisone at a dose of 5 mg daily or more or equivalent doses of other glucocorticoids, according to the FRAX criteria. Clinical information, such as LTOT, was obtained from the patient's medical records.

This study was approved by the Institutional Review Board of Keio University School of Medicine (UMIN Clinical Trials Registry, UMIN000001273), and all patients granted their written, informed consent to participate.

Pulmonary function tests

Vital capacity (VC) and FEV₁ were measured, using an MFR-8200 electronic spirometer (Nihon Koden, Tokyo, Japan). Diffusing capacity for carbon monoxide per alveolar volume (D_LCO/VA) was estimated by a 10-s breath holding (Chestac-55V; Chest, Tokyo, Japan).

Pulmonary emphysema score

Pulmonary emphysematous lesions were identified as low attenuation areas (LAA) on computed tomography scan (ProSeed, GE Healthcare Japan, Tokyo, Japan). To evaluate the emphysematous changes according to the Goddard classification (9), the whole lung was divided into left and right upper, middle and lower lung fields. LAA were visually scored from 0 to 4 in each division, where grade 0 indicated no LAA, grade 1 indicated LAA in 1–25%, grade 2 in 26–50%, grade 3 in 51–75%, and grade 4 in 76–100% of the lung area under examination. Two pulmonologists unaware of the patient's clinical status graded the areas, and assigned a total score between 0 and 24, providing a quantitative assessment of the emphysematous changes.

Bone status

BMD of both femoral heads and of the 2^nd, 3^rd and 4^th lumbar vertebrae was measured on a Lunar Prodigy® Advance, dual energy X-ray absorptiometric scan (GE Healthcare Japan). The diagnosis of osteoporosis was based on the lowest T-score at the 3 measured locations, and defined according to the WHO criteria, where osteoporosis is defined as a T-score < -2.5, osteopenia as a T-score ≥ -2.5 and < -1.0, and normal bone as a T-score ≥ -1.0.

Radiographs of the thoracic and lumbar spines were obtained, using a KYO-80S digital X-ray system (TOSHIBA Medical Systems, Otawara, Japan). We measured the height of the anterior and posterior borders and the mid segment of the vertebral bodies from T4 to L4. Fractures were diagnosed according to the criteria set for primary osteoporosis by the Japanese Society for Bone and Mineral Research: a) the height of the anterior border is >25% lower than the posterior border; b) the height of the mid segment is >20% lower than the anterior or posterior borders; or, c) the height of all 3 vertebral measurements is >20% lower than the measurements made on the adjacent superior and inferior vertebrae.

Serum and urine biomarkers of bone turnover

The serum concentrations of bone formation markers, uncarboxylated osteocalcin (ucOC) and a bone alkaline phosphatase (BAP), were measured, using, respectively, an electro-chemiluminescence immunoassay and a chemiluminescent enzyme immunoassay. The urinary concentrations of N-telopeptide of type 1 collagen (NTX) were measured as a bone resorption marker with an enzyme-linked immunosorbent assay. The cut-off concentrations were 4.5 ng/ml for serum ucOC¹⁰, and 29.0 U/l for serum BAP and 54.3 nmol bone collagen equivalents (BCE)/mmol creatinine for urinary NTX according to the Japanese guideline for prevention and treatment of osteoporosis 2006. The lower limits of detection for ucOC and BAP in serum and urinary NTX were 0.39 ng/mL, 2.0 U/L, and 1 nmol BCE/L, respectively.

Statistical analysis

We used a single and multiple variable logistic regression analysis to identify factors potentially associated with a risk of osteoporosis or vertebral fractures. Variables associated with P values < 0.1 in the single variable regression analysis were entered in the multiple variable regression model. The 10-year probability of suffering a major osteoporotic fracture was calculated with the Japanese version of the FRAX tool (http://www.shef.ac.uk/FRAX). We used one-way analysis of variance to compare the VC and FRAX scores versus the number (0, 1, ≥2) of vertebral bone fractures. Odds ratios (OR) and 95% confidence intervals (CI) were calculated. The statistical analyses were performed, using the StatView® software, version 5.0 (SAS Institute, Cary, NC). P values < 0.05 were considered statistically significant.

Results

The characteristics of the 85 study participants, who presented with moderate to very severe COPD, and who were not treated for osteoporosis, are shown in Table 1. According to the GOLD 2009 classification, 23 patients (27%) were in stage II, 30 (35%) were in stage III, and 32 (38%) were in stage IV, including 25 patients treated with LTOT for a median duration of 31 months (range 1–132).

Table 1.  Characteristics of the 85 study participants

Men	78 (92)
Age, years
< 65	10 (12)
65–74	31 (36)
≥ 75	44 (52)
Body mass index
< 18.5	10 (12)
18.5–24.9	61 (72)
≥ 25.0	14 (16)
Current smoker	10 (12)
> 60 g/day alcohol consumption	8 (9)
Medical history
Rheumatoid arthritis	3 (4)
Fractures	29 (34)
Parental hip fracture	7 (8)
Drug therapy
Corticosteroids	9 (11)
Orala
Inhaledb, μg/day
0	44 (53)
200–799	19 (23)
≥ 800	20 (24)
Warfarinb	14 (16)
Oral corticosteroidsa or warfarinb	22 (26)
Years of oxygen therapy
< 2	12 (14)
≥ 2	13 (15)
Pulmonary function tests,%predicted
Vital capacity
≥ 90	42 (49)
80–89	16 (19)
< 80	27 (32)
Forced expiratory volume in 1 sec
≥ 50	27 (32)
30–49	45 (53)
< 30	13 (15)
DLCO/VA,c
≥ 50	16 (24)
30–49	28 (42)
< 30	23 (34)

Values are numbers (%) of observations.

^aCurrent or previous use, b current use, c n = 67 measurements

DLCO/VA: diffusing capacity of carbon monoxide per liter alveolar volume

An abnormally low BMD (T score < -1.0) was detected in 39 of the 85 patients (46%), of whom 19 (22%) had a diagnosis of osteopenia and 20 (24%) had a diagnosis of osteoporosis. The factors correlated with osteoporosis by single variable logistic regression analysis are shown in Table 2A.

Table 2.  Factors correlated with osteoporosis and vertebral fractures by single variable logistic regression analysis

	A. Osteoporosis		B. Vertebral fracture
	Odds ratio [95% CI]	P	Odds ratio [95% CI]	P
Female gender	27.40 [3.05–26]	0.003	1.41 [0.29–6.80]	ns
Age ≥75 years	2.03 [0.72–5.76]	ns	2.68 [1.05–6.86]	0.04
Body mass index
< 18.5	9.53 [2.14–42.40]	0.003	1.18 [0.30–4.64]	ns
18.5–25	1.00	-	1.00	-
> 25	0.31 [0.03–2.64]	ns	0.71 [0.20–2.53]	ns
Dyspnea scale ≥2	3.20 [1.01–10.2]	0.04	0.69 [0.47–3.13]	ns
Oral corticosteroids a	0.91 [0.17–4.75]	ns	3.62 [0.80–16.5]	0.09
Inhaled corticosteroids b, μg/day
0	1.00	-	1.00	-
200–600	0.73 [0.17–3.06]	ns	0.42 [0.16–2.13]	ns
≥ 800	2.59 [0.81–8.24]	ns	2.30 [0.76–6.94]	ns
Treatment with warfarin b	0.84 [0.21–3.47]	ns	2.98 [0.92–9.68]	0.06
Treatment with oral corticosteroids a or warfarin b	0.94 [0.30–2.99]	ns	4.57 [1.60–13.1]	0.004
Years of oxygen therapy
none	1.00	-	1.00	-
< 2	1.49 [0.34–6.40]	ns	1.38 [0.36–5.33]	ns
≥ 2	3.82 [1.07–13.6]	0.03	3.86 [1.10–13.50]	0.03
Vital capacity (% predicted)
≥ 90	1.00	-	1.00	-
80–90	1.05 [0.18–6.09]	ns	1.80 [0.52–6.22]	ns
< 80	6.87 [2.06–22.8]	0.001	3.00 [1.05–8.58]	0.04
Forced expiratory volume in 1 sec (% predicted)
≥ 50	1.00	-	1.00	-
30–50	1.10 [0.32–3.70]	ns	1.27 [0.43–3.71]	ns
< 30	3.77 [0.88–16.2]	0.07	5.02 [1.24–23.8]	0.02
Carbon monoxide diffusing capacity/liter alveolar volume (% predicted)
≥ 50	1.00	-	1.00	-
30–50	1.17 [0.19–7.20]	ns	0.93 [0.24–3.54]	ns
< 30	5.39 [0.99–29.3]	0.05	1.35 [0.34–5.36]	ns
Emphysema score	2.71 [1.35–5.48]	0.005	0.90 [0.58–1.40]	ns
Serum concentrations
Uncarboxylated osteocalcin > 4.5 ng/ml	2.64 [0.94–7.48]	0.06	2.02 [0.77–5.32]	ns
Bone alkaline phosphatase > 29.0 U/l	4.00 [1.11–14.3]	0.03	1.49 [0.42–5.24]	ns
Urinary NTX > 54.3 nmol BCE/mmol creatinine	6.11 [1.52–24.6]	0.01	1.25 [0.32–4.87]	ns
T score
≥ -1	-	-	1.00	-
< -1 and ≥ -2.5	-	-	1.25 [0.41–3.78]	ns
< -2.5	-	-	3.39 [0.95–12.00]	0.06

^aCurrent or previous use; bcurrent use; CI = confidence interval; NTX = N-telopeptide of type 1 collagen; BCE = bone collagen equivalent.

Osteoporosis was correlated with female gender (P = 0.003), a BMI < 18.5 (P = 0.003), MMRC dyspnea scale ≥ 2 (P = 0.04), LTOT for ≥ 2 years (P = 0.03),%predicted VC < 80% (P = 0.001), emphysema score (P = 0.005), serum BAP concentrations > 29.0 U/l (P = 0.03), and urinary NTX concentrations > 54.3 nmolBCE/mmol creatinine (P = 0.01). Weaker correlations (P < 0.1) were found between osteoporosis and a%predicted FEV₁ < 30%,%predicted D_LCO/VA < 30%, and serum concentrations of ucOC > 4.5 ng/ml. We found no correlation between a) low BMD and b) treatment with warfarin or corticosteroids, current smoking, alcohol consumption > 60 g/day, FEV₁/FVC, or serum concentrations of albumin, calcium, magnesium, or inorganic phosphate. By multiple variable logistic regression analysis, osteoporosis was correlated with female gender, low BMI, low%predicted VC, and an elevated serum BAP concentrations (Table 3A).

Table 3.  Independent predictors of osteoporosis and vertebral fracture by multiple variable logistic regression analysis

A. Osteoporosis	Odds ratio [95% CI]	P
Female gender	17.2 [1.35–219]	0.02
Body mass index < 18.5	7.41 [0.98–56.1]	0.05
Vital capacity < 80%	6.68 [1.66–26.8]	0.007
Serum bone alkaline phosphatase > 29.0 U/l	7.64 [1.51–38.6]	0.01
B. Vertebral fracture(s)	Odds ratio [95% CI]	P
Age ≥ 75 years	2.93 [1.01–8.47]	0.04
Treatment with oral corticosteroidsa or warfarinb	6.00 [1.89–19.0]	0.002
Forced expiratory volume in 1 sec < 30%	7.52 [1.77–32.0]	0.006
C. Multiple vertebral fractures	Odds ratio [95% CI]	P
Treatment with oral corticosteroidsa or warfarinb	11.4 [1.92–67.6]	0.007
Forced expiratory volume in 1 sec < 30%	8.24 [1.23–55.5]	0.03
LTOT ≥ 2 years	4.74 [0.82–27.5]	0.08

^aCurrent or previous use; bcurrent use, CI = confidence interval; LTOT = long-term oxygen therapy.

Vertebral fractures were detected in 29 of the 82 patients (35%) whose radiographs of the spine were interpretable. The factors correlated with vertebral fractures by single variable logistic regression analysis are shown in Table 2B. The presence of ≥ 1 fracture was correlated with age ≥ 75 years (P = 0.04), pharmacotherapy with oral corticosteroids or warfarin (p = 0.004), LTOT for ≥ 2 years (P = 0.03), < 80% predicted VC (P = 0.04), and < 30% predicted FEV₁ (P = 0.02). The relationship between vertebral fractures and presence of osteoporosis did not reach statistical significance (P = 0.06). By multiple variable logistic regression analysis, age ≥ 75 years, treatment with oral corticosteroids or warfarin, and a%predicted FEV₁ < 30% were independent predictors of vertebral fractures (Table 3B).

Fractures of ≥ 2 vertebrae likely to interfere with pulmonary function were identified in 10 patients (12%). Mean VC value was reduced with increasing number of fractures; 92.6 ± 2.2, 83.9 ± 3.4, 75.4 ± 6.2% of predicted values in patients with no vertebral fractures, 1 vertebral fracture, and multiple vertebral fractures, respectively (p < 0.005). Multiple vertebral fractures were also correlated with a) LTOT for ≥ 2 years (OR 6.00; 95% CI 1.27–28.4; P = 0.02), b) use of oral corticosteroids or warfarin (OR 5.70; 95% CI 4.42–22.8; P = 0.01), and weakly with c) < 30% predicted FEV₁ (OR 6.00; 95% CI 0.92–39.2; P = 0.06). Multiple variable logistic regression analysis also shows that these factors were independently associated with ≥ 2 vertebral fractures (Table 3C). The 10-year probability of suffering a major osteoporotic fracture, calculated with the FRAX algorithm, tended to be higher in patients with ≥ 2 vertebral fractures than in patients with 1 or no fracture, though the difference was not statistically significant (Figure 1).

Figure 1.  Relationship between a) the 10-year probability of major osteoporosis-related fractures calculated with the FRAX algorithm; and, b) the number of vertebral fractures detected in our study population. All between-groups differences are statistically non-significant.

Discussion

Our analyses showed that, besides the known risk factors, such as older age, treatment with oral corticosteroids, and low BMD, impaired pulmonary function and chronic respiratory failure requiring LTOT were correlated with osteoporotic vertebral fractures. Female gender, a low BMI, high concentrations of serum or urinary biomarkers of bone turnover, and pulmonary emphysema were correlated with a low BMD, though not with vertebral fractures.

An earlier study has suggested a correlation between a low BMD and the severity of COPD (11). In that study, the prevalence of osteoporosis was 0% in patients in GOLD stage II, 9.6% in GOLD stage III, and 17.9% in GOLD stage IV. Although we excluded the patients who were being treated for osteoporosis, the prevalence of patients with T-scores ≤ 2.5 in our study was even higher, i.e., 17.4%, 13.3%, and 37.5% in GOLD stages II, III, and IV, respectively. The absence of correlation between BMD and FEV₁/FVC, observed in several previous studies (12–14), is noteworthy. The negative correlation we found between BMD and severity of pulmonary emphysema on computed tomography scan, is also concordant with recently published reports (14,15).

Although the diagnosis and management of osteoporosis are considered essential in the prevention of fractures in older persons, with or without COPD, other determinants of bone strength, such as the distribution and size of bone mass, the micro-architecture of the bone, the mineral-to-matrix ratio, the condition of the collagens and micro-injuries must also be considered (16). BMD alone does not reliably predict fractures, especially in older men (17). A meta-analysis suggests that measurement of BMD can estimate the risk of fracture in a population, though cannot identify individuals who will suffer fractures (18).

Therefore, the computer-based FRAX algorithm was developed to evaluate the risk of fracture based on age, sex, BMD, personal history of fracture and parental history of hip fracture, current smoking, alcohol consumption, treatment with corticosteroid, and history of rheumatoid arthritis and other causes of secondary osteoporosis (7). The Japanese version of FRAX was developed from data collected in a Hiroshima population of 2,596 participants, whose mean age was 65 years and 69% were women (7).

The results of our study, which showed a correlation between vertebral fractures and age, BMD, and treatment with corticosteroids or warfarin in patients with COPD, are partially concordant with these reports. Because our study population included small proportions of women, current smokers, consumers of large amounts of alcohol, and patients suffering from rheumatoid arthritis, we could not precisely measure the contributions of these factors in the estimation of risk of fractures. However, our results may suggest that the FRAX algorithm does not reliably predict the development of osteoporotic vertebral fractures in patients presenting with more than moderately severe COPD, as their calculated probability did not agree with their actual incidence. It is particularly noteworthy that the two indices of severity of COPD, a low FEV₁ and the presence of chronic respiratory failure, were both identified as additional risk factors, suggesting that a COPD-specific algorithm would estimate the risk of fracture with greater accuracy.

It has been reported that a low BMD, a history of fractures, and high concentrations of serum BAP in women predict a high incidence of fractures and that, in absence of these three risk factors, the 10-year probability of fracture is low (19). We had hypothesized that the blood or urinary biomarkers of bone turnover, such as BAP, ucOC, and NTX would contribute to the prediction of risk of fracture as supplemental factors, besides BMD or FRAX scores. However, these biomarkers offered no additional information regarding the risk of fracture in our population, though we did find a correlation between concentrations of biomarkers and BMD, as previously reported in a general population (20–22).

There are several limitations in our study. The studied population is potentially biased; patients with advanced stage of COPD were enrolled due to the recruitment in a tertiary referral hospital. Post-menopausal women were likely to be excluded, since most of them had already been treated for osteoporosis. The cross-sectional study design, which examines the prevalence, but not the incidence of vertebral fractures, is another limitation of the present study. A future study with a longitudinal cohort would be necessary to assess the validity of FRAX as a tool to predict the incidence of vertebral fractures in COPD patients.

In conclusion, our observations suggest that patients suffering from moderate to very severe COPD present with unique risk factors for loss of BMD and osteoporotic fractures, such as pulmonary dysfunction and chronic hypoxia. A disease-specific algorithm to predict the probability of osteoporotic fractures in patients suffering from COPD would enhance their management.

Declaration of Interest statement

This study was supported in part by a grant from the Ministry of Health, Labor, and Welfare to the Respiratory Failure Research Group (K.A. and T.B.), and by a grant-in-aid for Endowed Asthma/COPD Research Department from Glaxo-Smith-Kline plc (K.A. and T.B.). The authors alone are responsible for the content and writing of the paper.

Acknowledgments

We would like to give special thanks to late Dr. Akitoshi Ishizaka, professor of Keio University School of Medicine, for his encouragement, support, and advices to this study.