Dear Editor,

We thank Drs. Fu and Sheng [1] for their interest in our article [2]. Many studies [3,4,5,5] have shown the long-term results of bisphosphonate (BP) therapy. However, such studies were based primarily on the averaged results and did not address BP-unresponsive patients.

The ultimate goal of osteoporotic treatment is the prevention of a fracture [5, 6]. As it is impossible to demonstrate fracture prevention on a case by case basis, the evaluation of osteoporotic treatment can be estimated by surrogate markers. Bone mineral density (BMD) is the most useful marker for long-term evaluation since correlations with fracture risk have been established [5, 6]. In our cases, bone turnover markers clearly decreased by BP therapy. However, there have been several cases where initially increased BMD values decreased over time [2 and unpublished data].

Fu et al. stated that we unreasonably selected patients who were treated with BPs for 2 years and who were deemed to show a poor response to BPs. This claim is erroneous. Most patients had undergone BP therapy for longer than 3 years, and the average BP treatment period was 5 years. We selected non-responders among poor-responders based on five measurements of BMD values over 2 years since the assessment of BMD twice might have caused minus bias. We believe that minus bias may be minimized. Although, it is unknown that five measurements over 2 years are ideal.

We selected non-responders at the lumbar spine and hip and separated them from responders [2]. BMD values greatly increased in those non-responders. However, it remains unknown whether the BMD-increasing effects in non-responders are superior to those in BP responders.

As Fu et al. suggested, atypical femoral fractures (AFF) may increase. However, the frequency of AFF is quite rare, even with the use of BPs or denosumab [5, 7]. Also, there is no definitive evidence that osteonecrosis of the jaws (ONJ) is brought about by antiresorptive drugs [, 8]. We have published that patients with self-reported kyphosis had a significantly higher risk of problematic delayed wound healing after tooth extraction as compared with those without. Therefore, patients with vertebral fracture associated with osteoporosis may be at a greater risk for ONJ [9]. We maintain that the benefits of denosumab treatment are greater than the risks of ONJ or AFF even after BP therapy.

Adler et al. have recently described the ASBMR task force guidelines for BP therapy duration from a risk-benefit perspective [5]. Briefly, patients with major osteoporotic fractures or at high risk of fracture should continue BP therapy orally for up to 10 years [5]. Elsewhere, Taguchi et al. have reported that the frequency of ONJ and fracture increases after the cessation of BPs [10].

In conclusion, although this reply letter to our manuscript provides good suggestions on future studies in denosumab therapy, we believe that our currently published findings remain of value in terms of the benefits of denosumab for BP-unresponsive patients.