Low Bone Turnover and Microdamage? How and Where to Assess It?

It has been shown in a beagle dog model that suppression of bone turnover with bisphosphonates (BPs) is associated with an accumulation of microdamage in the spine, ribs, and iliac crest. Because these experiments used animals that were not estrogen deficient and did not have low BMD, it has been unclear whether similar accumulation would occur in postmenopausal women taking bisphosphonates. The paper by Chapurlat et al. reported the results from a study of iliac crest biopsies from 50 postmenopausal osteoporotic women who had received bisphosphonate treatment for at least 3 yr, with an average treatment duration of 6.5 yr. Comparing these specimens to biopsies obtained from 12 cadavers, they found a nonsignificant 2.5fold increase in microcrack density (p 0.59). They also reported that 54% of the women treated with BPs had no observable microcracks compared with 58% of the nontreated women. They conclude that, in “osteoporotic women on long-term BPs, microcrack frequency in the iliac bone is low, despite marked reduction in turnover.” Studies of microcrack accumulation in women treated with BPs are challenging because they rely on the collection of iliac crest biopsies that are difficult to obtain. Moreover, the quality of these specimens may not render them suitable for analysis of microdamage in every case. In addition, it is always difficult to obtain control samples; these generally would come from an established and approved clinical trial, because otherwise there is no indication to take a transiliac bone biopsy from a healthy woman who is not being treated with BPs. Notwithstanding these difficulties, there are several aspects of the work of Chapurlat et al. that make the conclusions of the study problematic. The study was carried out using only 12 control specimens, and the control sample was incompletely characterized. The controls were on average 16 yr older than the patient sample (p < 0.001). It is evident from numerous studies that the accumulation of microdamage is an agerelated phenomenon and that the exponential increase in accumulation begins at about 70 yr of age (i.e., older than the average age of the treated patients). Given the controls were obtained from the autopsy room, it is unclear why more precise age-matching was not achieved. Moreover, there was no medical history for the control sample, and therefore, it cannot be ascertained whether any or all of the controls may have been taking a BP, which BP they may have been taking, or for how long. This, in itself, makes the comparison with the patient sample invalid. Also, the study is underpowered, which the authors themselves point out. Twenty control samples and 80 treated patients would be necessary to achieve 80% power. Understandably the authors could not perform an a priori power calculation because there were no existing reports in the literature in which microcracks had been analyzed in human iliac crest biopsies. However, they could easily have done a power analysis midstudy, and because the control samples came from the autopsy room, there is no reason the population of controls could not have been expanded. The authors acknowledge this limitation, but state in their Discussion “The sample size is adequate to detect a sizable difference between the control and the treated groups but does not allow to capture [sic] a small difference of limited or no clinical relevance.” This statement is entirely unsubstantiated because no one knows what level of microdamage may be clinically relevant. The authors indicate that there was no significant association between bone turnover and microdamage accumulation. However, the control sample did not receive tetracycline labels, so activation frequency (Ac.f) could not be calculated. This restricts their regression analyses of microdamage to Ac.f to only those women treated with BPs, likely giving them a narrow range of turnover values (because all these women had low turnover) and making it unlikely that they would be able to find an association, even if one existed. It has been reported repeatedly that an exponential relationship exists between damage accumulation and Ac.f in beagle dogs, in which there is a wide range of turnover values. Even within the BP-treated group, the authors point out that double labels in 33% of the sample were obscure and could not be measured. Therefore, Ac.f measurements could only have been carried out on ∼34 of those women whose turnover rates were high enough to be able to detect the double label. These would also be the women less likely to accumulate microcracks because their remodeling rates were higher than those in which only single labels could be detected. There are also some more minor discrepancies in the paper as it appears on-line. For example, the treated and control samples are reversed in Fig. 2; the treated women had more than twice as much microdamage as the controls, but this figure shows the opposite. Figure 3 shows only 32 data points, but the treated sample consisted of 50 individuals. In Fig. 4, there are only 28 data points. Clearly, some of the data have been left out in these figures. The results of another recent study tend to contradict the conclusions of the Chapurlat et al. study. Stepan et al. measured transiliac crest biopsies from 38 women who had been treated with alendronate for an average of 5 yr and compared them to a control sample of 28 women who were treatment naïve. Although this study was not able to definitively conclude that suppression of bone remodeling in postmenopausal women treated with alendronate results in JOURNAL OF BONE AND MINERAL RESEARCH Volume 23, Number 7, 2008 Published online on March 10, 2008; doi: 10.1359/JBMR.080307 © 2008 American Society for Bone and Mineral Research