Neurosurgical approaches perform close to the occipital nerves are more likely to injure the GON and cause postoperative cranial pain [22, 23]. Some of these complications seem to be avoidable by good knowledge of the nerve anatomy of this region, careful positioning of the patient in the operating room, modification of the incision (risk of direct nerve damage) or modification of retractors placement (risk of indirect nerve damage).
The course of the GON within the dorsal cranio-cervical regions has been described in multiple anatomical studies [1–3, 5, 8–12]. As suggested by both the literature and our study, its course is relatively constant within the different muscular layers of this region. Conversely, the GON becomes subcutaneous after it crosses the trapezius, whose penetration location is variable. This anatomical inconsistency is confirmed by our results: even if most of the works in the literature stipulate that the nerve crosses the aponeurosis of the trapezius muscle rather than the muscle itself [8–10, 12, 14], our data show a muscle crossing in most of the cases (75%). The location of the exit point towards the subcutaneous region found in our study is consistent with the data in the literature [12]. However, the inter-individual variations observed in our dissections (10mm variation from the midline, 6mm variation caudal to the EOP) reinforce the fact that the subcutaneous part of the nerve course is subject to anatomical variability. We also found a close anatomical relationship between the GON and the OA, as described in the literature [1, 5, 12, 14]. The contribution of this anatomical knowledge allows us to understand that there is an increased risk of injury to the greater occipital nerve in the case of an approach starting with an incision situated between 1 and 4 cm lateral to the midline. However, if this type of approach is performed, careful dissection of the subcutaneous tissue will allow the identification of the GON, whose point of emergence is situated below the nuchal line. This identification will be facilitated by the identification of the OA.
Patient installation and positioning in the operating room also seems to influence the risk of developing postoperative occipital neuralgia. Four types of positioning are classically described for posterior fossa approaches [24] : prone position, lateral decubitus, supine position with head rotation, and sitting position, the latter being rarely used in current practice. Dynamic variation of the course of the GON during neck movements showed that extension of the neck do not affect tension on this nerve [4]. Conversely, head flexion movements appear to stretch the nerve between the obliquus capitis inferior and the semi spinalis capitis [4]. Rotational movements of the head cause a stretching of the contralateral C2 root [4]. Based on those considerations, there are a risk of GON injuries in the prone position and more specifically in the Concorde position (due to extreme head flexion) and in supine position with head rotation (due to stretching of controlateral C2 root).
Based on our results, the GON course and data from the literature [25], a midline approach should not cause direct GON injuries. Conversely, the hypothesis that the GON can be stretched and injured indirectly by the placement of retractors is a matter of debate. Because of its short distance to the midline (14mm), we noticed that the use of retractors to expose occipital bone can stretch the GON, thus suggesting that it can be injured indirectly. On the contrary, Sindou and Mertens studied the GON anatomical landmarks and considered that the midline suboccipital approach allows the integrity of this nerve to be respected [25]. Skins incisions for midline suboccipital approach can obviously not be changed. However, our results show that the GON stretching is proportional to the traction force exerted by the retractors. Furthermore, the placement of the retractors under the periosteum decreases the deformity exerted on the GON. Thus, assuming that there is a risk of indirect GON injury in this approach, we recommend placing the retractors under the periosteum and making a minimal retraction, just sufficient to expose the occipital bone.
There are multiple anatomical landmarks for the skin incision of the lateral suboccipital approach. In all cases, it is located between the midline and the retroauricular region and is therefore at risk of crossing the GON. In our study, we did not find any direct lesion of the GON, but its proximity to the incision (6 mm) and its anatomical variability lead us to believe that this risk does exist. Moreover, because of its close relationship with the incision path, there is also a risk of indirect GON injury with retractors placement. The risk of developing postoperative occipital neuralgia after surgery with a lateral suboccipital approach is poorly described in the literature. Most of the time, the approach described using the term lateral suboccipital approach is a retrosigmoid approach [26]. According to our results, this approach is nevertheless the most likely to cause GON injuries, and in particular the only one that could directly injure it. To minimize this risk, we recommend careful dissection of the subcutaneous tissue during this surgical approach. We also recommend placing the retractors under the periosteum, and achieving a minimal retraction, just sufficient for bone exposure.
Multiple incisions are described to perform a retrosigmoid approach. The most common are linear incisions medial to the mastoid and C-shaped incisions [27]. Most anatomical studies estimate risk of GON injury in this approach [27–29]. Aihara et al compared a S-shaped incision (where the GON was rarely cut but often stretched and retracted) and a C-shaped incision (where the GON was neither cut nor stretched during retraction) to explain the factors influencing the duration of postoperative headache in patients undergoing retrosigmoid approach for vestibular schwannoma [28]. They found a significantly shorter duration of postoperative headache in C-shaped incisions, suggesting that stretching of this nerve may lead to postoperative pain. Silvermann et al have described modifications to the retrosigmoid approach, including a skin incision that avoids the occipital nerves and a small craniectomy [29]. The combination of these two factors reduces the probability of occipital nerve injury and may help to reduce the incidence of postoperative headaches. According to the data of Sindou and Mertens [25], only a short oblique retromastoid incision (linear or curved), lateral enough to avoid the area where the nerve goes around the obliquus capitis inferior should not damage the GON. Those data seem to be confirmed by the increased risk of neck pain and headache after vestibular schwannoma removal via the retrosigmoid approach compared to the translabyrinthine [22, 23] and subtemporal [23] approaches. In our study, the most lateral point of the GON was 29 mm from the incision, which leads us to believe that direct injury to the GON is unlikely. Furthermore, no significant stretching of the nerve was found after traction on the various threads apposed along the approach, suggesting a very low risk of indirect nerve injury during the retrosigmoid approach. The discrepancy between our data and the literature seems to be explained by the fact that the incision we made for the retrosigmoid approach is very close to the ear. This is of course depending on the location of the sigmoid sinus. Thus, in order to minimize the risk of damage to the GON during the retrosigmoid approach, we recommend a fairly lateral approach if possible, 1 fingerbreadth medial to the mastoid. This incision should not extend too far caudally to avoid the point where the GON goes around the obliquus capitis inferior.
The semantics of the petrosal approach have been simplified by Miller who divides them into 2 distinct categories [30] : anterior and posterior petrectomy. Due to the preauricular skin incision in case of anterior petrectomy [30], only the posterior petrectomy approach may provide GON injury. For posterior petrectomy, an inverted J-shaped (or C-shaped) incision with a retroauricular starting point is classically performed [30]. In our study, we did not find any risk of injury to the GON with the petrosal approach. It is lateral enough to avoid the nerve, and no retraction towards the nerve is required. Our results are clinically confirmed by literature, where few postoperative neuralgias are found after petrosal approach. This risk appears indeed to be less frequent than for retrosimoid approaches [22, 23]. For example, a recent review of the literature lists 3 studies showing no significant difference in terms of pain between the retrosigmoid and trans-labyrinthine approaches and 4 studies showing a lower rate of postoperative headaches in patients who underwent petrosal approaches [31].
Our anatomical study applied to posterior cranial fossa approaches provides interesting data regarding the risk of intraoperative nerve injury. Performing initially a descriptive anatomical study of the GON allowed us to understand its course. The fact that we performed most of the approaches used in current practice and studied their relationship with the GON seems to us to be a methodological strength to meet the aim of our study. Thus, based on our study and the data in the literature, we propose advice to minimize the risk of intraoperative GON injury (table 4).
However, we admit several limitations to our study. First of all, there are multiple causes for postoperative occipital neuralgia, such as irritation of sensitive nerve branches for the occipital dura mater and the venous sinuses [32], neurogenic inflammation by irritation of the trigeminal nerve [22], muscular adhesions with the dura mater in the case of craniectomy [22, 26], muscular damage during the surgical approach [22] or muscular tension due to intraoperative positioning [22]. These different causes have not been considered in our work and may act as confounding factors. In addition, the GON course presents an inter- and intra-individual variability, and we only performed every approach twice. Similarly, we did not study the dynamic variations of the GON related to the positioning of the patient in the operating room. Our protocol for identifying the risk of nerve damage appears to be rigorous: traction was performed in the direction of the nerve, with an identical force each time. On the other hand, the assessment of the neurological risk seems to us to be more approximate, as it was only assessed on the deformation and stretching of the nerve. We also believe that this deformation may be overestimated, since dissection of the nerve frees it from its subcutaneous and muscular adhesions, thus favoring its mobility. Finally, the absence of the possibility of revealing clinical damage felt by the patient prevents us from estimating the clinical translation of our suppositions.