The prevalence of osteoporosis in COPD is higher than in healthy age-matched control subjects [3]. Moreover, COPD patients have a 60 to 70% higher risk of death following hip fracture than people without COPD [4]. It is therefore of high clinical importance to diagnose and treat osteoporosis in COPD according to international guidelines [2].

Medical history, current medication use, and smoking status were assessed. Post-bronchodilator pulmonary function tests were performed (Jaeger MASTERLABBODY, VIASYS Healthcare) and COPD was diagnosed and classified according to the GOLD guidelines [1]. Body mass index (BMI) was defined as low (<21kg/m²), normal (21–25kg/m²), overweight (>25–30kg/m²), and obese (>30kg/m²). Fat-free mass index (FFMI) was assessed by bio-impedance analysis (BODYSTAT 1500 medical, single frequency; Xitron Technologies) and classified as depleted (men <16kg/m² and women <15kg/m²) or normal [13]. A six-minute walking distance test was performed to estimate functional exercise capacity [14]. Arterial blood gases were collected to determine pH, arterial carbon dioxide tension (PaCO₂), and arterial oxygen tension (PaO₂).

Whole-body BMD as well as BMD at the femur and the lumbar spine (L1–L4) were assessed for each patient using a DXA-scan (Lunar Prodigy Ge-Lunar). Because the risk of hip and vertebral fractures is similar in men and women for any given BMD, we used the NHANES III data to calculate T-scores for both men and women [15, 16]. For whole-body DXA osteoporosis was defined by a T-score <−2.35, osteopenia: T-score between −2.35 and −0.9, and normal BMD: T-score >−0.9 [17]. For local DXA, osteoporosis was defined by a T-score of ≤−2.5, osteopenia as T-score between −1 and −2.5, and normal BMD: T-score >−1 [2]. To compare different locations of DXA-scan, we divided groups into osteoporosis and no osteoporosis (osteopenia and normal BMD combined) because only patients with osteoporosis need to be treated pharmacologically.

Twenty-five percent of the 130 patients with a T-score >−2.35 as determined by whole-body DXA did have osteoporosis based on local DXA. In contrast, only 3.7% of the 103 patients with a T-score >−2.5 as determined by local DXA did have osteoporosis based on whole-body DXA (Table 2). This difference in prevalence was highly significant (P < .0001). The relative risk of having osteoporosis on local DXA in case of no osteoporosis on whole-body DXA was 6.4 times higher than the risk of having osteoporosis on whole-body DXA in case of no osteoporosis on local DXA (Table 2).

Almost 16% of 122 patients with T-scores >−2.5 as determined by DXA at the lumbar spine did have osteoporosis based on DXA at the hip. In addition, a little over 18% of 126 patients with T-scores >−2.5 based on hip DXA-scanning did have osteoporosis on DXA-scan of the lumbar spine (Table 3). This difference was not significant (P = .644).

The prevalence of osteoporosis has been shown to differ when the same cut-off value for the T-score is used to define osteoporosis [18, 19]. This leads to an over- or underestimation of osteoporosis depending on the method used to asses BMD. In order to compare whole-body DXA to local DXA, we corrected for this difference in methodology by defining osteoporosis according to Boyanov [17] and the WHO [2], respectively. In turn, a good sensitivity-to-specificity ratio in the diagnosis of osteoporosis was obtained [17].

A possible explanation for the method-specific differences in prevalence of osteoporosis may be age-related. Indeed, loss of BMD of the lumbar spine occurs at a younger age than loss of BMD of the hip [6, 20]. In addition, due to degenerative changes of the spine at higher age, lumbar BMD could even be increased in the elderly. In the present trial the prevalence of osteoporosis was slightly higher using lumbar spine DXA (27.4%) compared to hip DXA (25.0%), while the prevalence of osteopenia was clearly higher in the latter group (33.3% versus 52.4%). Consequently, these patients may be at risk to develop osteoporosis.

The International Society for Clinical Densitometry stated that the BMD of the spine and the hip should be measured and diagnosis of osteoporosis should be based on the lowest T-score of either the spine or the hip [11]. Indeed, in our COPD patients we found the prevalence of osteoporosis based on DXA of the hip to be 25%, of the lumbar spine 27.4%, and of the hip and lumbar spine combined 38.7% (Table 1). In combination with the fact that 25% of patients without osteoporosis on whole-body DXA had osteoporosis on local DXA, it seems reasonable to advise DXA of the hip and the lumbar spine and diagnose osteoporosis based on the lowest T-score in COPD patients. The costs of a whole-body DXA are the same as a local DXA scan. In addition, the costs of a DXA of the hip or the lumbar spine only are equal to the costs of a DXA scan of both the hip and lumbar spine. Therefore, there are no economical reasons to limit DXA-scan to the hip only. In addition, the time to make a local DXA scan is less than the time to make a whole body DXA. More research is needed to assess risk factors for osteoporosis in COPD in order to determine which patients should be evaluated for osteoporosis.

The current study showed that the group of patients without osteoporosis on whole-body DXA but with osteoporosis on local DXA consisted of significantly more females than males, patients with a lower six minutes walking distance and more patients with GOLD IV and higher PaCO₂ and lower PaO₂ levels as compared to patients without osteoporosis on both whole body- and local DXA (Table 4). Therefore, in this group of patients, it seems even more important to use local DXA to determine BMD instead of whole-body DXA. Even more because in the general population the fracture risk is increased in females [21] and in COPD patients, a higher prevalence of osteoporosis has been found in patients with higher GOLD classification and/or hypercapnic patients [22–26]. In the current study, the increased number of patients with GOLD IV COPD in group II is probably not due to increased airway obstruction (FEV1 was not significantly different in both groups), but to hypercapnia and/or hypoxemia (Table 4). Indeed, Dimai and colleagues found a significantly lower BMD and increased serum cross-linked telopeptide of type I collagen (a bone resorption marker) in hypercapnic patients suggesting that hypercapnia induces increased bone resorption [23]. With decreased daily physical activity, hip BMD especially will be decreased, which will be compensated by the BMD of other non-weight baring body parts when whole body BMD is assessed.

When results of whole body DXA-scanning were added to local DXA-scanning, only 5 extra patients without osteoporosis on local DXA had a T-score of <−2.35 on the whole body DXA-scan (2.4%).