In the present study, the bone microarchitecture of the axis was analyzed using ex vivo HR-pQCT. The results were correlated with CT data from patients to derive implications for the occurrence and care of axis fractures. The major findings were: (1) CT-based apparent densities in zone I were higher than in zone II and III, mainly independent of fracture occurrence. (2) Cortical and trabecular microarchitecture parameters decreased from zone I in the tip of the DAX to zone III in the CAX, while trabecular separation was lowest and trabecular number highest at the apex accordingly. (3) The trabecular and cortical tissue mineral density was similar at the base and tip, while lowest at the CAX. (4) The SMI indicated a plate-like and dense, sclerotic trabecular microarchitecture in the tip of the DAX transforming into a highly cross-linked rod-like trabecular microarchitecture in the CAX.

In clinical cases, we demonstrated lower apparent density in zone II and III than in zone I, irrespective of fracture status. In addition, the sagittal plane of zone II in the fracture group, but not the non-fracture group, showed a higher apparent density than the sagittal plane in zone III. This difference could be explained by the higher apparent density in the fracture area due to the occurrence of hematoma and the impaction of trabeculae. Although no differences in individual zones were detected between fractured and non-fractured patients, there was a pronounced regional heterogeneity in both the non-fracture and the fracture group with lower apparent density in zones II and III compared to zone I. This could be a correlate of reduced bone quality at the base of the DAX and in the CAX.

As it is known that clinical CT scans provide only a rough estimate of BMD, since the apparent density in HU cannot be directly translated into BMD, the bone quality and microarchitecture of the axis was further analyzed experimentally using HR-pQCT on full bone specimens. To our best knowledge, this is the very first study to analyze the bone microarchitecture of the axis via HR-pQCT in a comparable sample size of n = 28, following a clinically relevant fracture classification and correlating the findings to clinical groups of patients with and without fractures of the axis. We suggest that the decreasing cortical and trabecular microarchitecture from the tip of the DAX to the CAX is a major factor influencing fracture susceptibility. Notably, similar analyses have been previously performed by our group in other skeletal regions such as the distal fibula, inferring fracture mechanisms based on regional heterogeneity in local bone microarchitecture [24].

In 1994, Amling et al. analyzed the microarchitecture of the axis histologically in n = 22 autopsy specimens with a mean age of 50 years [15]. For analysis, they also chose a division of the axis into three zones following the classification of Anderson and D'Alonzo. They found a BV/TV of 20% in the CAX, 10% in the base of the DAX, and 26% in the odontoid process. The BV/TV was significantly lower in the base of DAX compared to the other zones. Moreover, the trabecular pattern factor, a parameter that indirectly accounts for inter-trabecular connections by determining the relation of convex and concave surface patterns [25], was determined in these regions, with the worst trabecular connection detected at the base of the DAX. The group concluded the presence of a region of least resistance in the base of the DAX due to lower bone mass and weaker bone microarchitecture [15]. In a subsequent study, Amling et al. provided additional data from n = 11 autopsy specimens with known osteoporosis and deduced an increased risk for fractures and subsequent non-unions of the dens in the osteoporotic bone due to impaired bone quality in the base of the DAX. Although providing important insights, these previous analyses were performed only two-dimensionally in the sagittal plane [15, 16]. Our data offer insights into the bone quality and microarchitecture in the corresponding regions of the axis three-dimensionally. From the current data, it can be concluded that in zone II, which corresponds to the base of the DAX, and in zone III in the CAX, which may include a residual subdental synchondrosis, the trabecular bone is weaker than in zone I in the apex of the DAX due to a lower bone mass with less and thinner trabeculae. Further, the differences in BMD indicated higher mineralization within the two cranial zones supporting this assumption. Regarding the microarchitecture of the trabecular bone, the SMI and the Conn.D indicated a dense, sclerotic, and plate-like structure at the tip of the dens, while the CAX presented with high interconnectivity.

Previous studies have shown that local bone mass and microarchitecture have a major influence on the occurrence of fractures [26,27,28,29,30,31]. By analyzing bone specimens from lumbar vertebrae via image-guided failure assessment under microtomographic imaging, it was shown that a decrease in BV/TV is accompanied by an increased fracture probability and that especially the combination of low BV/TV and Conn.D indicates the weakest bone region. The authors concluded that the weakest region within a bone structure may be crucial for fracturing the whole complex and that the above mentioned structural parameters are crucial for identifying these weak regions [26]. Furthermore, a positive correlation of a low Tb.N and a high Tb.Sp with the occurrence of vertebral fractures has been shown in the past [30]. Although no biomechanical studies with combined consideration of high-resolution microarchitecture of the axis have been performed to date, previous observations suggest that microarchitecture plays a major role in influencing the occurrence of fractures. However, the occurrence of fractures of the axis must be considered in a more differentiated manner and cannot be attributed to a single aspect such as bone quality. Both anatomical variations, i.e., bone structure, bone composition, joints, and ligaments, and trauma mechanism are decisive. A recent biomechanical study investigated the influence of bone density on the occurrence of axis fractures indicating an increased fracture risk with low BMD while the direction of the applied loading showed little influence [31]. Nevertheless, especially the articular connection between the anterior arch of the atlas and the DAX seems likely to be relevant for the occurrence of the DFTII. While fractures of the axis in young patients often arise in the context of high-energy traumas, a typical trauma mechanism in geriatric patients is a frontal head impact causing a reclination in the upper cervical spine. It is likely that this anatomical feature and its resulting high mechanical moments accompanied by a low BMD in the base of the DAX determines the high fracture susceptibility for DFTII.

Our data show that in the extension of the DAX caudally into the CAX, Ct.Th, BV/TV, and trabecular as well as cortical TMD continued to decrease, whereas Amling et al. postulated an increase in BV/TV [15]. These differences might be due to methodological discrepancies (i.e., 2D vs. 3D analysis) and age differences of the groups investigated. Due to the obvious difficulty in obtaining samples, there are no studies that have investigated the trabecular bone of the axis using HR-pQCT, in a similarly large collective to date. Recently, Wang et al. analyzed n = 5 dry bone samples with a mean age of 52 years [23]. Four volumes of interest (VOI) were analyzed within the trabecular bone of the DAX. VOI I was defined in the tip of the DAX, VOI II in the neck of the DAX, VOI III in the body of the DAX, and VOI IV in the base of the DAX. In this previous study, a higher bone mass, and thicker and more numerous trabeculae were determined in VOI I compared to VOI IV, whereas trabecular separation was lower in VOI I than in VOI IV [23]. While these findings obtained in a small collective of n = 5 specimens are generally consistent with our data, an analysis of CAX was not performed in the study by Wang et al. [23]. In our view, an additional analysis of CAX seems essential. The classification most commonly used in clinical practice to classify fractures of the axis, regardless of the gaps that certainly exist, is the classification according to Anderson and D'Alonzo [1]. DFTII followed by DFTIII occur most frequently, whereas DFTI are very rare [2]. While DFTII occur at the base of the DAX, DFTIII occur within the cylinder that results from an extension of the DAX caudally into the CAX. Furthermore, in this cylinder lies the screw trajectory for anterior screw fixation, frequently performed for the surgical treatment of a DFTII, according to Böhler et al.[11]. For these reasons, we considered an analysis of the bone microarchitecture in zones I–III based on the classification of Anderson and D'Alonzo to be suitable, to derive clinical implications.

The following clinical implications emerge from our and previous results: The base of the DAX is prone to fracture based on the combination of low bone mass, bone mineralization, and trabecular microarchitecture [15,16,17, 20, 23]. In addition, the subdental trabecular bone represents a biomechanical weak point within the axis. On the one hand, this may account for the frequent occurrence of DFTIII in addition to DFTII. On the other hand, our data provide a possible explanation as to why the typical cut-out of screws after anterior screw fixation often occurs in zones II and III, which was also supported by the data described by the CT-based apparent densities in the clinical part of the study. Therefore, we consider anterior screw fixation to bear a high risk for failure, especially in geriatric patients [9, 12, 13]. However, due to the important advantages of the technique compared to alternative or non-surgical procedures, such as the low invasiveness and the preserved rotational ability between atlas and axis, we consider the establishment of alternative osteosynthesis procedures for the placement of lag screws in the DAX with a separate anchorage in the CAX, such as osteosynthesis plates, essential to increase the quality of care and safety of elderly patients with fractures of the axis. These aspects should be further investigated in additional studies.

Limitations of our study include that only non-fractured axis specimens could be analyzed by HR-pQCT. Thus, no direct comparison of microarchitecture between non-fracture and fracture groups could be made, which should be performed in the future. Although full autopsy allowed us to exclude conditions that locally affect skeletal microarchitecture (e.g., tumors), the presence of osteoporosis could not be determined by established methods. Nevertheless, the autopsy specimens offered us a unique opportunity to perform the high-resolution HR-pQCT examination, which is usually limited to distal bones (radius, tibia) [32] and cannot be performed on the cervical spine in the clinical setting. Another limitation is that HR-pQCT measurements of the specimens were performed with a resolution of nearly 15–30% (depending on the region) of the trabecular thickness. Hence, the partial volume effect might have an influence on the mineralization results of this study. However, additional calculation showed similar correlation coefficients between TMD and thickness in the thinnest and thickest structures, indicating a negligible influence (p = 0.76). Another limitation of our study is that the analysis was limited to a geriatric patient population. The resulting narrow age range prevented a meaningful analysis of age-related bone loss, and thus the presumed influence of age on fracture risk or screw loosening.

In conclusion, the axis is characterized by a decreasing cortical and trabecular microarchitecture from the tip of the DAX to the CAX, indicating inferior bone quality in this region. These findings may partly explain the clinical observation that zones II and III, the base of the DAX and in the CAX, depicture sites with a higher fracture susceptibility than zone I, the tip of the DAX. While non-bony anatomical structures and trauma mechanism certainly also play a critical role regarding fracture occurrence, the reduced bone quality in zones II and III could be a risk factor for implant loosening after anterior screw fixation. In order to ensure safe osteosynthesis of axis fractures, improved, additional anchorage of the implants in zones II or III, e.g., by osteosynthesis plates, might be indicated, especially in aged patients with limited bone status.