Abstract
It has long been postulated that antiresorptive therapy may improve the quality of life of patients with spinal metastases by lowering the risk of vertebral compression fracture (VCF) induced or exacerbated by stereotactic body radiotherapy (SBRT).1-3 Recently we read with great interest the article by Patel et al 1 entitled “Antiresorptive Medications Prior to Stereotactic Body Radiotherapy for Spinal Metastasis are Associated with Reduced Incidence of Vertebral Body Compression Fracture”. The authors had investigated the potential effects of such antiresorptive agents as denosumab and bisphosphonate in the largest cohort of spinal metastasis hitherto. They concluded that patients benefited from antiresorptive medication initiated prior to SBRT in terms of protecting against VCF. Specifically, compared to no usage at all, antiresorptive administration before SBRT and throughout decreased the incidence of VCF (1-year post-SBRT: 4% vs 12%, P = .045). As indicated by Figure 1, however, another apparent yet seemingly paradoxical finding—spinal metastasis patients receiving antiresorptive treatment only initiating after SBRT even had a higher risk of VCF in comparison to those taking no antiresorptive medication at all—was barely reported or discussed in the article.
The authors conducted a Kaplan-Meier survival analysis of the VCF incidence against the antiresorptive use status. With three groups denoted by colorful curves, the main results were straightforwardly presented in Figure 1. Nevertheless, several statistical details were not clarified, which might hinder the interpretation and understanding. Firstly, whether the P value of .008 indicated an overall comparison result between three groups or a post-hoc pairwise test (antiresorptives before SBRT vs no antiresorptives)? If the latter, whether a raw or modified P value was reported, and which procedure, for example, the Bonferroni correction, was adopted to account for multiple testing? Otherwise, taking the VCF incidence at 1-year post-SBRT for example, we could not tell whether the difference is statistically significant (if a modified P = .045 < .05) or not (if a raw P = .045 > .05/3). Besides, the statistical software utilized by the authors was not reported anywhere.
The authors seemed to have overlooked a paradox: pre-SBRT-initiating antiresorptive usage lowered the VCF incidence when compared with no antiresorptive usage at all; meanwhile, in comparison to no usage, post-SBRT-initiating antiresorptive usage alone seemed to increase the VCF risk, particularly in the first year of follow-up. A higher cumulative incidence of VCF up to 1-year post-SBRT was evidently demonstrated by the green curve (no antiresorptive usage, 12%) positioned higher than the blue one (pre-SBRT-initiating antiresorptive usage, 4%) in Figure 1. Likewise, the orange curve (post-SBRT-initiating antiresorptive usage, ∼18%) extended higher above the green one (no antiresorptive usage, 12%) up to 1-year post-SBRT, exactly indicating the paradox of “less is more”. The VCF incidence in the group of antiresorptive usage only after SBRT might have been unintendedly under-reported. Instead, the authors only mentioned the comparisons between two other groups (antiresorptives before SBRT vs no antiresorptives) in the second paragraph of the Results part.
To lower VCF risks in spinal metastasis patients planning to receive SBRP, whether and when to initiate the antiresorptive treatment to account for both clinical efficacy and cost-effectiveness is of great significance.4,5 We hold that the authors could check the statistical issues concerned to clarify the paradox and draw a robust conclusion.
