J Bone Miner Res. 2026 Feb 23:zjag039. doi: 10.1093/jbmr/zjag039. Online ahead of print.
ABSTRACT
It is essential to closely monitor the response to oral bisphosphonate therapy for osteoporosis, as many patients are nonadherent. The conventional approach is to monitor whether changes in BMD or bone turnover markers exceed the least significant change. That approach assumed that if a patient were not receiving a treatment, there would be no change in the BMD. We propose an alternative approach, that of expected change. We define this expected change as the change in bone mineral density (BMD) (at 24 months) or bone turnover markers (at three months) that is exceeded in 90% of patients who are adherent with oral bisphosphonate therapy. We studied 108 postmenopausal women (age<85 years) who were randomised to the licensed dose of alendronate, ibandronate or risedronate treatment for two years, along with calcium and vitamin D supplementation. We identified the performance of BMD and bone turnover markers in three ways. We calculated the signal-to-noise ratio, which was lower for BMD [4.1 and 2.1 for lumbar spine BMD (LSBMD) and total hip BMD (THBMD), respectively] compared to bone turnover markers (9.4 and 10.2 for CTX and PINP, respectively). We estimated the response rate as the percentage of women exceeding the least significant change, which was lower for BMD (47% and 24% for LSBMD and THBMD, respectively) than for bone turnover markers (96% and 94% for CTX and PINP, respectively). We estimated the expected change as the 90th (or 10th) percentile of change in adherent patients. We required the expected change to exceed the least significant change, and this was not observed for LS- and THBMD, but it was observed for CTX and PINP (expected changes of 0.233 ng/mL and 12.1 ng/mL, respectively). Thus, the bone turnover markers CTX and PINP showed the best performance as response markers for monitoring oral bisphosphonate treatment, and the new approach is based on a biological rather than a statistical endpoint when using the expected change approach.
PMID:41725388 | DOI:10.1093/jbmr/zjag039
