Does Denosumab Really Improve Muscle Strength? Current Evidence Is Weak
Article information
Dear Editor,
We read with interest the recent systematic review and meta-analysis published by Aryana et al.1) entitled “Denosumab's therapeutic effect for future osteosarcopenia therapy: a systematic review and meta-analysis.” The meta-analysis included four studies and concluded that denosumab (a human monoclonal antibody osteoporosis medication) improved hand grip strength (but not lumbar spine bone mineral density or gait speed) to a greater extent than bisphosphonates (another antiresorptive bone medication).
We wish to draw attention to several methodological flaws that are likely to have influenced the outcomes and interpretation of this meta-analysis.
(1) Different denosumab group comparisons were combined in the meta-analyses. For example, Fig. 3 illustrates a comparison of mean change in hand grip strength between denosumab users and bisphosphonate users from Miedany et al.2) and a comparison between denosumab users and no therapy from the other two studies (Bonnet et al.3) and Rupp et al.4)). Furthermore, the Bonnet hand grip strength data in the meta-analysis are incorrect because corrected hand grip strength data published in a 2020 corrigendum5) to the Bonnet et al.3) paper were not considered. In fact, mean change in hand grip strength in the denosumab group was closer to 0 kg than 3 kg, and mean hand grip strength change in the bisphosphonate group was approximately -2 kg.
(2) Data with different units of measurement were included in the meta-analyses. For example, the lumbar spine data shown in Fig. 2 represent a combination of absolute change (Bonnet et al.3)), percentage change (Rupp et al.4)), and T-scores (Miedany et al.2)). For mean change in hand grip strength, the Figure 3 forest plot represents a combination of absolute change (Bonnet et al.3) and Miedany et al.2)) and percentage change (Rupp et al.4)).
(3) The authors calculated mean change from baseline in Figs. 2 and 3 for Meidany et al.2) but appear to have simply used follow-up standard deviations instead of calculating the missing standard deviations for the change to follow-up.
(4) The absence of a “leave-one-out” sensitivity analysis makes it difficult to ascertain how individual studies contributed to the overall effect size estimate.6)
(5) The Hartung-Knapp-Sidik-Jonkman method7) for random effects meta-analyses should have been applied to reflect uncertainty in the estimation of between-study heterogeneity due to the small number of studies included.
(6) The authors state that “…funnel plots of all analyses…did not reveal any significant asymmetry.” However, tests for funnel plot asymmetry are not recommended for meta-analyses including <10 studies as the test power is too low to differentiate real asymmetry from chance.6)
The studies included in the meta-analysis also had notable shortcomings. None of the studies was randomised and controlled to the standard of a clinical trial. One was a non-randomised longitudinal study in which participants were classified into active (denosumab) or control (bisphosphonate) groups based on prescribed medication,2) one was a retrospective observational study with propensity score matching,4) one was a prospective observational study with participant matching,3) and one, reported in the format of a letter to the editor, included data from a longitudinal cohort study.8) We note our previous letter to the editor in Osteoporosis International9) that raised concerns about the methods of Rupp et al.4) including the influence of outliers and the lack of adjustment for group differences on their findings. The Bonnet et al. study3) included interesting animal data (for which a 2023 corrigendum was also recently published10)), however, the modest human data extracted from that study is lacking the basic elements of a rigorous clinical trial design. Miedany et al.2) did not report participant age, 65 of their participants were ‘contaminated’ with prior bisphosphonate exposure of unspecified duration, and Tables 2 and 3 contain group mean scores for many outcome measures that are unusually similar and, in many cases, identical. Importantly, none of the included studies controlled for the confounding effect of exercise. In fact, changes in strength and/or function from baseline of any patient group who have received medical advice (as all participants presumably had) could easily reflect simultaneous advice to undertake exercise (per best practice recommendations). Thus, in the absence of a genuine control group, monitoring of exercise exposure with a validated instrument, and subsequent control for physical activity uptake in each study, it is not possible to attribute any functional changes to medications that were simultaneously prescribed.
Considering the above concerns with the methodology of the meta-analysis and notable limitations of the included studies, the conclusion of Aryana et al.1) that denosumab “… may be favoured in individuals with osteosarcopenia to improve muscular performance and reduce falls risk” based on a single dubious finding of greater hand grip strength improvements in denosumab users than bisphosphonate users is questionable.
Therefore, it is premature to suggest that denosumab use per se will enhance muscle strength. The promotion of such an idea would undoubtedly be at the expense of the one therapy known to enhance muscle strength (resistance training), which would be a disservice to patients considering therapy to prevent osteoporotic fractures.
Notes
CONFLICT OF INTEREST
The authors claim no conflicts of interest.
FUNDING
None.
AUTHOR CONTRIBUTIONS
Conceptualization, BRB, DS, RMD, JK; Writing-original draft, BRB; Writing-review & editing, BRB, DS, RMD, JK.