Classification of the healing pattern
Fifty-two cases (36 male and 16 female) were reviewed, with a mean age of 41.5 years. The median post-operative follow-up was 3.2 years (from 12 months to 74 months). The baseline characteristics are shown in Table 1. CT/MR scans taken at the final follow up were carefully analyzed to evaluate the healing status. All of the fractured vertebrae obtained bony union after surgery, and there was one case of instrumental failure. However, cavities at the healed vertebra body were seen in 31 cases (59.6%). According to the morphology of the healing vertebral body, the patients were divided into the Intact Vertebra group (n = 21) and the Vertebral Cavity group (n = 31). The former group had an intact vertebral body without any cavity, while the latter group had at least one recognizable cavity (diameter ≥ 3 mm) in the vertebral body. These two groups were further classified into four healing types based on the integrity status of the endplates or the morphology of the cavities (Figure 1).
The Intact Vertebra group (Type I/II)
Twelve cases (23.1% of total cohort) in the Intact Vertebra group had intact endplates and were classified as Type I, the Complete Healing type (Figure 2). In the cases of this type, both vertebra and endplate were completely healed without any cavity or splitting.
Nine patients (17.3% of total cohort) in the Intact Vertebra group had small cavities at the endplate region and were therefore classified as Type II, the Endplate Cavity type (Figure 3). This type showed bone healing in the major part of the vertebral body but had a small defect or cavity in endplate. Interestingly, nearly all of the patients in type II had cavities at the middle or anterior part of the upper endplate.
The Vertebral Cavity group (Type III/IV)
Nineteen patients (36.5% of total cohort) in the Vertebral Cavity group were classified as Type III, the Spherical Cavity type, as they had single spherical cavities in the vertebra (Figure 4). The cavity in most of these cases was located in the anterior, middle, or upper part of the vertebral body, in which endplates were usually corrupted. The cavity was usually wrapped with a hardened layer, which may imply that the cavity was difficult to heal further.
Twelve patients (23.1% of total cohort) with burst vertebra in the Vertebral Cavity group were classified as Type IV, the Burst Cavity type (Figure 5). All of the type A4 fractures in this cohort were manifested as type IV healing. Most of the burst vertebral bodies were connected, with either upper or lower endplate broken. Only a few cases showed burst throughout the entire vertebral body, with both upper and lower endplates broken. The multiple fracture fragments are incompletely healed with wide splitting lines (> 1 mm) and peripheral sclerosis. The cavities were located in the center part of the vertebral body, manifested as an intersection of several splitting lines.
Risk factors for vertebral cavity
In order to investigate the risk factors for the formation of vertebral cavities, demographic data between the Intact Vertebra group and the Vertebral Cavity group were compared. We found the proportion of males in the Intact Vertebra group (47.6%) was significantly lower than in the Vertebral Cavity group (83.9%, P < 0.05). There were more patients with a history of smoking in the Vertebral Cavity group (P < 0.05). Vertebral cavities were seen in all types of A4 burst fracture cases, and the median ASIA impairment scale of the Vertebral Cavity group was significantly worse than that of the Intact Vertebra group (P=0.029). These results indicated that the formation of vertebral cavities is related to gender, smoking history and the severity of the fracture. No significant difference was observed in age, location of the fracture, or the completion of spinal canal decompression between the two groups.
Influence on clinical outcomes
Clinical outcomes were also compared to evaluate the clinical significance of vertebral cavities. The VAS score of back pain at the last follow-up did not showed significant difference between the Intact Vertebra group and the Vertebral Cavity group (with an average VAS of 0.57 point and 1.07 point, respectively), implying that most vertebral cavities were painless. There was no significant difference between ODI outcomes of patients with or without vertebral cavities (Table 2). The median ASIA impairment scale of the Vertebral Cavity group was still worse than that of the Intact Vertebra group at the time of follow-up, although statistically insignificant (P=0.101). Our results suggested that most of the vertebral cavities were asymptomatic, but their impact on neurological and functional outcomes were still unclear.