Reply to: Grade 1 vertebral fractures, bone mineral density, and the risk of fractures

Abstract

To the Editor We appreciate that Dr. E. Kanterewicz et al. have shown interest in our study, in which we investigated associations between grade 1 vertebral fracture (VF), verified by vertebral fracture assessment (VFA), and incident fractures.(1) In their Letter, the authors point out that prior to our study, they had reported that VFA-verified prevalent minor vertebral deformities (MVDs) were associated with an increased risk of incident VFA-verified VF.(2) They also refer to other publications about associations between prevalent VF and incident VF, diagnosed by VFA.(3-5) Indeed, all these reports and our recent study to some extent support the usefulness of VFA to detect VF, and the importance of prevalent VF, including grade 1 VF or MVD, in assessing fracture risk in postmenopausal women. Although there are similarities between previous studies and our study, there are important differences. Kanterewicz et al.(2) examined a considerably younger age group (59–70 years) than the population of older women (75–80 years) included in our study. This is reflected by the large difference in VF prevalence at study inclusion (24.2% compared to 4.1%). In the study by Kanterewicz et al.,(2) both a morphometric classification and the Genant methods were used to identify VF, in contrast to our study, which solely utilized the Genant method, making it difficult to compare the results. It is also a fair assumption, that in the age group 75 to 80 years, identifying mild VFs is more challenging because of commonly present osteoarthritis, scoliosis, and sometimes impaired image quality, due to positioning difficulties, compared to younger age groups. As pointed out, by Kanterewicz et al. in their Letter, the study from Kadowaki et al.(3) found that prevalent vertebral deformities in middle-aged Japanese women were associated with a threefold increased risk for subsequent VF, but the prevalent vertebral deformities were not graded. Thus, the impact of mild vertebral deformities was not addressed, disabling a meaningful comparison with the results obtained in our study.(3) Ferrar et al.(4) graded prevalent VFs but could not find a significant association between prevalent mild VF and incident VF after adjustment for age and total hip bone mineral density (BMD). They also concluded that among postmenopausal women without prevalent VF but with short vertebral height (SVH) there was no association between SVH and incident VF.(4) Interestingly, this result contradicts the findings by Kanterewicz et al.(2) that MVD (vertebral height ratios between –2 and –2.99 standard deviations [SDs] below the reference) confers an increased risk of subsequent VF. Greendale et al.(5) found that two-thirds of the prevalent vertebral deformities were grade 1. They investigated the incidence rate of vertebral deformities but failed to detect a predictive role of grade 1 vertebral deformity on incident fracture.(5) However, the investigated population was considerably younger, and both the prevalence and incidence of VF much lower than in our study, arguably reducing the relevance of the comparison. Kanterewicz et al. further commented that the associations between grade 1 VFs, incident any fracture and incident VF, were dependent on BMD in our study. In fact, the association with incident any fracture (adjusting for all risk factors and BMD) was borderline significant (hazard ratio [HR] of 1.51; 95% confidence interval [CI], 0.98–2.34) and for incident VF, the risk estimates highly similar, but with much wider 95% CI (HR 1.52; 95% CI, 0.71–3.25). Our interpretation is that these results do not support lack of independent association between grade 1 VF and fracture incidence, but rather of the opposite, with the nonsignificant result being due to the low number of fractures and the resulting insufficient statistical power in the analyses, especially regarding incident VFs. Kanterewicz et al. argue that the incidences of VF in our study and in theirs were very similar. We disagree. Kanterewicz et al.(2) used follow-up scans to identifying incident VFs in their cohort, in contrast to our study, in which we only used digital x-rays to verify clinical fractures, which have a much lower incidence than morphometric VFs. Thus, the incidence ratios for VFs cannot be compared between studies. Finally, we can agree with Kanterewicz et al., and as stated in the introduction of our publication, that the literature is inconclusive regarding the role of grade 1 VF in predicting future fragility fractures. We believe that our study and those discussed herein contribute to the understanding of the clinical importance of grade 1 VFs. The authors have no conflicts of interest or relationships to disclose. Lisa Johansson: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; software; validation; writing-original draft; writing-review & editing. Mattias Lorentzon: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; resources; software; supervision; validation; writing-original draft; writing-review & editing.