KD is a common complication of VCFs after conservative treatment. Epidemiological studies estimated that 7–37% of patients with VCFs may progress KD. Despite its prevalence, the exact etiology of KD remains unclear, although vertebral ischemic osteonecrosis is the leading hypothesis supported by recent studies [20]. Anamnestically, many patients with KD experience a prolonged pain-free symptomatic period following initial trauma, only to present later with recurrent back pain. At this stage, radiographic signs such as IVC and vertebral collapses with kyphotic deformities are often observed. Identifying associated factors for KD is crucial for risk stratification. While not all patients need immediate surgery, high-risk individuals should undergo regular follow-ups. Timely interventions, including surgical augmentation, should be considered if disease progression occurs. In our study, we retrospectively reviewed the personal medical history, epidemiological characteristics, and imaging data of 66 KD patients. We then selected 104 patients with old vertebral compression fractures as the control group. Finally, six independent predictors associated with KD were identified: age, BMD, history of osteoporosis, vertebral compression ratio, vertebral compression morphology, and disc compression grade.

Several studies support our findings. For example, KO et al. [1] reported that the prevalence of KD is increasing in aging populations, with a strong correlation between low BMD and the occurrence of IVC in middle-aged and elderly VCF patients. Our results also highlight that the risk of KD increases significantly with age ≥ 70.5 years, and BMD (T-score) ≤ −3.65. Wu et al. [11] noted that over 80% of IVC cases occurred at the thoracolumbar junction, which is consistent with our study. This may be attributed to the range of motion and spinal loading intensity and greater susceptibility to sagittal imbalance experienced at the thoracolumbar junction [21, 22]. However, we did not find the thoracolumbar junction location to be an independent associated factor for KD, possibly due to sampling variability or inherent susceptibility of this area to fractures.

Importantly, our study found that patients with co-existing osteoporosis were more likely to progress to KD following vertebral compression fractures. Findings from previous studies regarding that osteoporosis disrupts bone homeostasis, which can lead to vertebral microfractures, hematomas, and subsequent ischemic cycles that may result in vertebral collapse [23, 24]. Furthermore, the anatomy of vertebral arteries, particularly the lack of collateral blood supply to the anterior third of the vertebral body, might explain why wedge-shaped fractures are more likely to develop into KD [25]. Additionally, although using the average height of adjacent vertebrae above and below the compressed vertebra is a common method [13], it may not be suitable for our study. Firstly, in patients with multiple compression or pathological changes in adjacent vertebrae, this approach could lead to inaccuracies. Secondly, the natural height variation at the thoracolumbar junction, where KD predominantly occurs, makes it challenging to accurately reflect normal vertebral height using adjacent vertebrae. Therefore, we believe the formula 1-AVH/PVH × 100% is more appropriate for calculating vertebral height loss in our study. Ultimately, Our analysis supports the association between higher vertebral compression ratios (≥ 29.9%) and an increased likelihood of KD progression.

Additionally, our study innovatively explored imaging predictors such as vertebral compression ratio, compression morphology, Cobb angle, and disc degeneration—factors often overlooked in earlier studies. Among them, the effect of disc degeneration was first proposed by our team. The intervertebral disc, as a compartment of the functional unit of the spine, plays an important role in transmitting and absorbing mechanical stresses on the spine and in maintaining normal vertebral body movement [26]. Both endplate calcification and nucleus pulposus dehydration are further exacerbated as the disc state deteriorates, and disc degeneration leads to a further deepening of the ischemic and hypoxic state, which ultimately results in disc cell death and a decrease in disc height, leading to an increase in regional kyphosis and a concentration of anterior stress [27, 28]. In addition, disc degeneration leads to decreased cushioning capacity and increased mechanical stress, which promotes intervertebral instability. Moreover, relevant studies have also shown that disc degeneration is accompanied by the production of inflammatory factors [29], and the presence of inflammatory mediators and inflammatory cells has been similarly found in KD [30]. Therefore, in order to investigate the association between disc degeneration and the occurrence of KD, we did the subsequent analysis and found that there was a statistical difference between the two, especially when the disc degeneration reached grade III or above, the incidence of KD would be significantly increased.

Contrary to some literature, our study did not establish a significant correlation between the history of glucocorticoid use or long-term alcohol consumption and the onset of KD. Zhang et al. reported that long-term glucocorticoid therapy may elevate the risk of developing KD [31]. Other studies have identified glucocorticoids as a key predisposing factor for IVC and vertebral nonunion [1, 10, 32]. Additionally, long-term alcohol consumption was reported to increase the risk of KD, which may be related to the increase of micro-fatty emboli in the terminal arteries due to alcohol [3]. The correlation between these two factors and KD is related to ischemic osteonecrosis [33]. However, other studies indicated that osteonecrosis is multifactorial, with risks from alcohol and glucocorticoids being time and dose dependent, possibly mixed with other factors [34]. Our initial findings aligned with this, but after excluding other associated factors, these two predictors were not independent, suggesting they may play a role alongside other associated factors.

In summary, our study provides valuable insights into the predictive factors for KD but also has limitations. First, this single-center, retrospective study had a relatively small sample size. Further multicenter, prospective studies are necessary to validate our findings. Second, the short follow-up period may limit the detection of disease progression in the control group, necessitating longer follow-up. What’s more, because the sample selection for this study relied on patients recorded at our hospital, there was an inherent selection bias in retrospective studies and was not representative of all KD patients, especially those with mild symptoms. Furthermore, the selection of some data, such as bone mineral density, was based on the initial visit, which may introduce potential bias. Finally, future studies should aim to investigate additional associated factors that may contribute to the onset of KD.