Anterior cervical disc fusion is nowadays daily performed in every spine surgery center to treat cervicobrachial neuralgias resulting from degenerative osteophytes arising from the cervical spine through the intervertebral foramina. On one hand, the safety of this procedure has been widely demonstrated, even for multilevel surgery, so that it can be sometimes routinely performed in ambulatory surgery centers [11–13]. On the other hand, compression of nerve roots can be multifactorial in degenerative cervical pathologies, with bone spur formation, reduced disc height and foraminal stenosis narrowing the intervertebral foramen, thus making surgery more difficult. Even if it may be useful in every procedure, the role of indirect decompression is particularly important in those cases, where direct removal of pathological bone can be associated with several complications, such as for instance vertebral artery injury [14].
Indirect decompression can be defined as the decompression of neural tissues without removing the compressive tissue [15]. In anterior spinal procedures, it is mainly performed by the distraction of the intersomatic space and restoration of disc height with interbody implant or graft, which lead to an opening of the intervertebral foramen [7, 15]. Most of the studies focusing on indirect decompression in spinal surgery concerns the lumbar spine, since significant increases in foraminal dimensions have been demonstrated after Anterior Lumbar Interbody Fusion (ALIF) [16–18] and after Oblique Lumbar Interbody Fusion (OLIF) procedures [19]. As a consequence, there is a lack of data concerning aspects of decompression in cervical spine surgery and factors that could influence it.
Our results show a statistically significant post-operative improvement in patient suffering cervicobrachial neuralgia resulting from osteophytic foraminal stenosis (p < 0,001) with 85% of them having a post-operative enhancement in all the questionnaires filled, and with none of them having suffered post-operative complication. This is consistent with the literature [10, 11, 13, 20], thus reinforcing the fact that ACDF is a safe and efficient procedure for the treatment of selected cervical degenerative pathologies. Among age, ASA score, BMI, operative time and smoking status, smoking was the only individual variable that had an impact on patient reported outcome, with smokers having statistically lower post-operative NDI scores (p:0,032). This reinforces the already known notion that smoking is a pre-operative predictor of a worse outcome after ACDF [21, 22].
Pre-operative radiological measurements were consistent with the values expected, with disc height, foraminal height and foraminal volume being lower than the ones of asymptomatic patients [23]. Furthermore, foraminal volume was found to be greater in level C6-C7 than in superior levels, being also consistent with the reported literature [23]. A significant increase in disc height (p < 0,001), foraminal height (p < 0,001) and foraminal volume (p < 0,001) was showed post operatively, with an increase of every parameters measured at every level that were fused. Increase in disc height was significantly associated with increase in foraminal height (p < 0,001) and foraminal volume (p < 0,001), thus highlighting the fact that loss of cervical disc height is a component of the foraminal volume reduction leading to cervical radicular pain. What's more increase in foraminal height was significantly correlated with foraminal volume (p < 0,001), and was the major component affecting post-operative increase in foraminal volume. Considering all those informations, it seems clear that gain of disc height with implants allows foraminal height to be greater, thus increasing foraminal volume. Previous studies have shown that disc space distraction with graft or implant insertion lead to an increase in foraminal height [7, 22], foraminal anteroposterior diameter [5] or foraminal area [5, 9, 22] after ACDF, and more especially if the bone graft is placed as posterior as possible [10]. Even if compression of the nerve roots frequently occur at the entrance zone of the intervertebral foramina [24], we consider the fact that the nerve roots can be compressed anywhere while it passes through the foramen, and thus we believe that foraminal volume measurement might be more predictive of a nerve roots compression.
There is a lack of data regarding clinical outcome and radiological parameters after ACDF. Suk et al found that post-operative foraminal dimension is negatively correlated with post-operative radicular pain (p:0,002) [10]. After statistical analysis, we did not find any association between increase in disc height, foraminal height or foraminal volume and patient reported outcome, except that increase in foraminal height was statistically associated with improved NDI score (p:0,037). However, even if non-statistically significant with all the scores, we tend to think our main hypothesis is valid. First of all, we found a correlation between gain of disc height and both increase in foraminal height and volume. Indirectly, improvement in NDI score with increase in foraminal height let us suppose that NDI is affected by both increases in disc height and foraminal volume. This is supported by our p value which are close to the significance level (0,11 for foraminal volume, 0,21 for disc height). Secondly, we found that patient reported outcome was worse with increase in disc height being less than 175% whereas a major improvement of all scores appears with increase in disc height higher than 175% (Table 4). Thirdly, foraminal decompression should free the nerve roots, thus resulting in a more marked improvement in NDI and radicular VAS scores than in cervical VAS scores. Our result show that p value is higher for VAS cervical pain than for NDI and VAS for radicular pain in every measured parameter (Table 4). Fourthly, our small number of included patients might favor sampling fluctuations, so that the reliability of our statistical analysis could be affected.
Another lack of published data concerns the optimal improvement of disc height needed while performing ACDF. In fact, in daily practice, most surgeons decide the height of the implant based on intra operative findings and their own experience, and indirect decompression is neglected [7]. This attitude is not without consequences, with small graft or implants resulting in higher incidence of complications, and over distraction affecting the stress over the posterior column [25]. Wang et al showed that improvement rate was worse with restoration of disc height being less than 2mm or more than 6mm [26]. Suk et al propose a distraction of 2 to 3 mm to obtain a satisfactory clinical outcome [10]. Celik et al suggest that an oversized graft or cage might compromise the adjacent foramen [7]. Based on cadaveric studies, An et al found that ideal graft height was directly correlated with pre-operative disc height [27], and Yang et al set the cut-off 160% of improvement, with size of intervertebral foramen decreasing with distraction over that value [25]. Based on our result (Table 4), it seems (although not significant) that improvement in clinical outcome occurs with increase in disc height being more than 175% of the preoperative height, without major changes in questionnaires with a gain of more than 200%.
Regarding ACDF, our study provides update on surgical management of osteophytic foraminal stenosis. In the first description of ACDF by Robinson et al, uncovertebral osteophytic spurs were not removed [1], while a few years later Cloward emphasized that direct decompression of the unco-vertebral osteophytes had to be main part of this surgery [2]. Recent studies concluded that partial uncovertebral joint resection in ACDF does not provide significant increase in foraminal dimensions over ACDF alone [28]. Furthermore, Shen et al fount that unco-vertebral resection should not be systematically undertaken during ACDF because of similar outcomes and iatrogenic risks [20]. Those data suggest that focusing on indirect decompression might increase foraminal size and contribute to pain relief, while minimizing surgical iatrogenic risk. Even if Albert et al disprove the importance of indirect decompression, finding no significant correlation between graft height and change in foraminal dimension or pain relief, they agree that foraminal dimensions significantly increase after surgery in a few sample of foraminal dimension [9]. To our mind, graft height is not a good indicator of indirect decompression, which should rather be studied by calculating the percentage increase in disc height. With much more foramen measurements in our study, we assess that foraminal height and foraminal volume are significantly correlated with gain of disc height and play a part in clinical outcome after surgery.
Through its methodology, our study seems to have reliable and fair results. The use of self-administered questionnaires appears to be the best technique to compare pre- and post-operative data, while removing investigator variability. What's more, NDI score achieve a high degree of internal consistency and reliability [29]. The strengths and limitations of our method of volume calculations are given in a previous published study [23].
However, we admit some limitations in our study. First of all, the small number of patients included reduce the veracity of our statistical analysis and limit the extrapolability of our results. Secondly, only one third of patients underwent a single level surgery. Even if multilevel surgery increases the number of intervertebral foramens studied, it leads to difficulties in analyzing patient reported outcome because of impossibility to know exactly how each level interacts with the pain experienced. Thirdly, our follow-up time was quite short (3 months), which has 2 consequences. First, the natural course of symptoms regression can be longer than 3 months after ACDF. Secondly, we cannot evaluate the risk of pseudarthrosis, which could explain for example the fact that 3 patients had worse outcome at follow-up. Finally, foraminal size vary with cervical curvature [22] and movements of flexion and extension [6, 28], which were not considered in our study.