Response to the letter to the editor: Risk factors for cage subsidence following anterior–posterior spinal fixation in osteoporotic vertebral fractures: a multicenter retrospective study
Article information
, Shinji Takahashi,1, Hiroyuki Yasuda2, Masaki Terakawa2, Sadahiko Konishi2, Minori Kato1, Hiromitsu Toyoda1, Akinobu Suzuki1, Koji Tamai1, Akito Yabu1, Yuta Sawada1, Masayoshi Iwamae1, Yuki Okamura1, Yuto Kobayashi1, Masato Uematsu1, Hiroshi Taniwaki1, Hiroaki Nakamura1, Hidetomi Terai1
We thank the authors for their thoughtful comments regarding our study and for highlighting important considerations in interpreting our findings [1].
First, we agree that the number of outcome events was limited relative to the complexity of the multivariable model. Although cage subsidence occurred in 53 of the 100 patients, the model included five explanatory variables, with endplate injury represented by two indicator terms. In particular, no patient with grade-2 endplate injury was present in the non-subsidence group, resulting in sparse data and a wide confidence interval for this variable. Therefore, although the analysis suggests associations between cage subsidence and preoperative vertebral instability, low Hounsfield unit (HU) values, short-segment fixation, and endplate injury, the magnitude of the adjusted odds ratios should be interpreted cautiously. We consider these findings exploratory and hypothesis-generating rather than definitive estimates for individual risk prediction.
Second, we agree with the concern regarding the cutoff values for Δlocal kyphosis and HU. The thresholds of −14.0° and 87.5 HU were derived using receiver operating characteristic analysis and the Youden index and were subsequently evaluated in the same cohort. Consequently, their discriminatory performance and effect estimates may be optimistic. Our intention was to provide clinically interpretable candidate thresholds, not to establish universally applicable treatment criteria. These values should therefore not be adopted as definitive clinical cutoffs without validation. Future studies should evaluate Δlocal kyphosis and HU as continuous variables, assess potential nonlinear relationships, and validate the proposed thresholds using internal resampling and, preferably, an independent external cohort.
Third, we acknowledge the possibility of confounding by indication in the association between fixation range and cage subsidence. The extent of fixation was determined by each operating surgeon rather than by a standardized protocol. Thus, the choice of construct may have reflected clinical factors that were not fully captured in our dataset, including fracture morphology, vertebral instability, bone quality, adjacent-level conditions, and the surgeon’s assessment of mechanical failure risk. Adjustment for HU, Δlocal kyphosis, fracture level, and endplate injury cannot eliminate residual confounding arising from these unmeasured factors. Accordingly, our results do not establish a causal protective effect of longer fixation. Rather, they suggest that fixation range may be one component of risk and surgical planning that warrants further investigation. A prospective study with predefined indications for fixation length, or a sufficiently powered analysis using methods such as propensity-score adjustment, would be needed to clarify this relationship.
We appreciate the opportunity to clarify these issues. Our study identifies potentially important factors associated with cage subsidence, but external validation and more robust risk-modeling approaches are required before the reported estimates or thresholds can be used for definitive clinical decision-making.
Sincerely,
Yuki Kinoshita on behalf of the authors
Notes
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
Author Contributions
All authors conceived, drafted, critically revised, and approved the final version of this letter.