Sagittal area changes can be covered by endplate length and the posterior height of the vertebral body. In the early period, area was mainly influenced by the reduction in endplate length. As middle vertebra had two shortened endplates, its area change was more significant than that of superior and inferior vertebrae. At last follow-up, endplates were stable, whereas the posterior height of all three vertebrae increased, which may represent HO formation (taking into account the high prevalence of HO and its progression at last follow-up) and consequently led to an increase in the area (Fig. 2).

The anterior-posterior diameters can be classified into three groups: (1) For the four endplates that meet the artificial disc (e.g., inferior endplate of superior vertebra), their lengths significantly decreased early and remained static after. (2) The two endplates that do not meet the disc (e.g., superior endplate of superior vertebra) had a tendency to grow in length. (3) No significant change in waist lengths.

For group (1), the reduction in length can be explained by ABL. ABL was present in 60% of the segments in this study, and in approximately 25% of the segments, ABL involved both endplates. The change in group (1) is consistent with the bone resorption observed in previous studies on CDA [10, 13, 14] and is considered to be an adaption of bone to a new biomechanical environment [15]. According to Frost’s theory, resorption will occur to reduce the amount of unwanted bone when the functional load applied to the bone does not reach the desire [16]. A finite element analysis showed that the posterior part of the Prestige LP carried more pressure, which may imply a redistribution of pressures on endplate after implantation [17]. Group (3) did not seem to be involved in this process. Inside the vertebral body, superior endplate pressure may converge during downward conduction, pass through the waist region, and redistribute eventually to inferior endplate. Thus, regardless of changes in endplate pressure distribution, the sum of pressure converging on the waist remains the same and causes no bone reconstruction in this region. However, the above is only speculative and requires further validation by biomechanical studies.

Factors related to surgical manipulation may also contribute to ABL. The four endplates are subjected to surgical procedures such as burring. Due to their natural curvature, the endplates are prepared to better match the footprint. These insults could lead to bone remodeling and ABL. According to this explanation, bone resorption would not have occurred to the waist as it was barely surgically damaged, like what we observed.

It is worth noting that, as endplates untreated, group (2) exhibited opposed length changes to group (1). Possibly, this represented a continuation of natural degeneration accompanied by osteophytes formation at margin of the vertebral body [18]. Although surgery terminates the degeneration of operated segments, other segments remain exposed to factors associated with cervical spondylosis such as age-related changes, which may lead to continued degeneration [19].

Middle vertebra had significant early area changes and happened more ABL. This could be due to its unique position: between two surgical segments, middle vertebra endured a “double” injury; clamped by two prostheses, middle vertebral body experienced a “double” bone reconstruction. However, rigorously, the intra-operative preparation of superior and inferior endplates is not identical [14], and the bone reconstruction may differ, too. Therefore, the change may not be a simple doubling. Besides, the endplates on middle vertebra themselves could be more prone to ABL in early post-operation. Opposed to the results of Kieser et al. [14], we detected differences in the incidence and degree of ABL in endplates with different positions (“between implants” or “not between implants” [14]), but the differences did not last beyond 6 months after surgery. The different results could be due to different prosthesis type, because Prestige LP was not included in their study.

The exception of middle vertebra deserves attention. To minimize the risk of complications, the ABL risk that middle vertebral body may suffer should be fully considered when planning multisegmental surgery and during intra-operative procedures, especially endplate milling and burring. Maximal preservation of middle vertebra volume can leave a buffer for post-operative ABL.

Lin et al. [20] observed collapse of the anterior edge of middle vertebral body in four patients after contiguous two-segmental ACDF with a zero-profile implant, while we did not observe a similar phenomenon in our study. They attributed the collapse to stress concentration and inner blood supply damage. The movable design of artificial disc may disperse the concentrated stress. In addition, in their study, four screws were inserted in middle vertebra. In contrast, the Prestige LP is stabilized by no screw but rails [21], which may cause less damage to the internal vertebral blood supply.

Overall and segment ROMs were well recovered and maintained. T1S varied between 22.53°±7.04° and 26.00°±6.37° over all follow-ups. SVA, although fluctuated, was not significantly changed, varying between 12.7 mm ± 6.9 and 15.0 mm ± 7.3 mm. These are consistent with our previous findings: CDA, although effective in preserving mobility, has a limited ability to improve sagittal alignment [22]. However, these parameters are still largely within the range to achieve good clinical outcomes according to previous studies. The review by Ling et al. suggested that the ranges are as follows: C7 or T1 slope with a mean value of 20° and no higher than 40°, cervical SVA with a mean value of 20 mm and no higher than 40 mm [23].

The present study has the following key learning points. In contiguous two-level CDA, the middle vertebral body is special. More ABL happened to it, leading to detectable morphological changes on sagittal plain radiograph. In addition, this study proposed a new parameter, vertebral body sagittal area. Combined with anterior-posterior diameters and vertebral body heights, it helped to quantify morphological changes. For example, there has been no consensus on the method of comparing the degree of ABL per vertebra (not only per endplate), but in our study, the differences were detected successfully by sagittal area.

This study has the following limitations. Preoperative osteopenia was demonstrated to cause a higher ABL incidence [24], but the few osteopenia cases in our sample restricted further analysis. HO was assessed only on lateral radiographs, where there was some difficulty in identifying due to facet joint overlapping [25]. Furthermore, the conclusions drawn from this study are specific only to the Prestige LP artificial disc and may not be generalizable across other types. Finally, this study was a retrospective study with variable follow-up times.