High variability of facial muscle innervation by facial nerve branches: A prospective electrostimulation study

To examine by intraoperative electric stimulation which peripheral facial nerve (FN) branches are functionally connected to which facial muscle functions.


INTRODUCTION
There have been many anatomical reports regarding the macroscopic and microscopic course of the peripheral facial nerve to the mimic muscles. 1,2 Typically, coming out of the stylomastoid foramen, the main trunk of the facial nerve is divided into a temporofacial division and a cervicofacial division. 3,4 The temporofacial division should have temporal, zygomatic and buccal branches. The cervicofacial division most frequently has a mandibular branch and a cervical branch. The allocation of these five main branches to the mimic muscles from anatomical textbooks is shown in Supporting Table 1 in the online version of this article. Actually, a clear anatomical allocation of main peripheral branches to individual mimic muscles during facial nerve dissection, for instance during parotidectomy, is often not possible because the peripheral branching and the intercommunication of the branches are highly variable. 3,5,6 The temporofacial division most often has a plexiform arrangement formed by dichotic and anastomotic divisions. The cervicofacial trunk resembles merely a simple large loop, but up to about 30% of its fibers show anastomoses with the temporofacial division. 7 In nearly half of the cases buccal branches arising from the two main divisions are interconnected with the zygomatic branch. 8 Even when preparing the facial nerve in an anterograde direction beginning at the main trunk and beyond the bifurcation into temporofacial and cervicofacial division, as it is mainly performed during parotidectomy, a clear assignment of the just-prepared nerve branch to the main branches (temporal, zygomatic, buccal, mandibular, cervical) might be difficult due to the variability of the nerve branching network. However, this is of importance because only such an assignment allows the surgeon to decide if facial nerve monitoring recording of only a limited number of muscles will track nerve stimulation and alert him before damaging the nerve. Facial nerve monitoring is still performed mostly with two-channel systems with recording from the orbicularis oris and orbicularis oculi muscle. 9 This means that only the facial nerve branches are monitored, which are functionally connected to these two muscles. Due to the above-mentioned reasons this can be very variable. Therefore, when always using the same standard recording positions, the quality of the monitoring is variable. Furthermore, for future bionic applications to restore facial nerve functions, it is necessary to know which nerve branches reveal which facial movement and with which variability. 10 Hence, the present prospective clinical study was performed to systematically stimulate the exposed facial nerve and its branches during parotidectomy to evaluate the alignment of the branches to functionally important facial movements.

Study Design and Setting
This prospective observational study was performed from June 2014 to December 2015 at the Department of Otorhinolaryngology, Jena University Hospital, Jena, Germany. Approval for the study was obtained through the local institutional review board, and informed consent was obtained from all study participants. Seven patients were investigated (four male, three female; median age 62 years). Standard lateral parotidectomy (three right side, four left side) was performed because of a benign tumor (four patients), malignant tumor (two patients), and chronic sialadenitis (one patient). Standard facial monitoring (CLEO Nerve Monitor; inomed, Emmendingen, Germany) with recording electrodes (SDN electrodes RT/SW and Trigon GN; inomed) in the frontal, orbicularis oculi, orbicularis oris, and depressor labii inferioris muscles was performed. Facial nerve function was normal before and after surgery in all cases.

Electrostimulation, Facial Monitoring, and Recording of Facial Movements
During parotidectomy, the main trunk of the facial nerve and all main branches were exposed under an operating microscope until the anterior border of the parotid gland. The peripheral facial nerve and its branches were systematically stimulated first with a monopolar electrode with ball tip (straight, length 5 13 cm, no. 525209; inomed) and then with a bipolar concentric stimulation electrode (BCS probe, length 5 13 cm, no. 522106; inomed) beginning with the main trunk, all branches of first order, and all branches of second order. The monopolar stimulation electrode has a larger stimulation field allowing a faster screening of the tissue for nerve branches. The bipolar electrode has a small stimulation field allowing a meticulous scanning of a once-discovered facial nerve branch. Each branch was stimulated from proximal to distal. The stimulation at each stimulation point started with 0.1 mA and was increased stepwise to 2 mA with a frequency of 3 Hz. Switching between the frequency of 3 Hz to 30 Hz was useful to verify the direction of the facial movements. The stimulation with 3 Hz evoked only short contractions of the subdermal muscles visible often only as a shiver of the skin. When stimulated with 30 Hz, a tetanic contraction of the whole stimulated muscle was visible, and the direction of the muscle movement was clear to see. Each stimulation was repeated at least one time. The exposed facial nerve and its branches and the whole electrostimulation procedure were video recorded via a microscope camera (Image 1 HD H3-M; Karl Storz, Tuttlingen, Germany). Simultaneously, the facial contractions were video recorded by an endoscope camera (Image 1 HD H3-Z; Karl Storz) fixed with a supporting arm over the face. Both camera signals were recorded simultaneously together with one audio channel for audio-comments into the same video documentation system (Advanced Image and Data Acquisition NEO; Karl Storz).

Evaluation of the Facial Movements
The evaluation was performed offline independently by two investigators. All facial movements were documented on a standardized electronic form. Depending on the individual facial nerve branching, the stimulation of seven to 12 branches was recorded and analyzed. To be able to compare the highly variable nerve branches and summarize the results of all patients, the branches of the facial nerve were numbered in a standardized manner (see Supporting Figure 1 in the online version of this article). The main trunk of the facial nerve was number 0. After bifurcation, the first cranial main branch (temporofacial division) was number 1, the first caudal main branch number (cervicofacial division) was number 2. A trifurcation was not seen. Following this nomenclature, we continued to number the branches of the second order starting from cranial with number 11, 12 for sub-branches of the first cranial main branch number 1. In the same manner, the numbers 21, 22, and so on from cranial to caudal were used for the sub-branches of the second cranial main branch number 2. Branches of third order were numbered 211, 212, and so on in the same order. If the pattern of facial muscle movements changed during the proximalto-distal stimulation of a branch, this was also classified adding a letter (a, b) to the number of the branch. For instance, the second branch of the temporofacial division (number 12) showed a different stimulation pattern during proximal stimulation (number 12a) than during distal stimulation (number 12b). The videos were evaluated according to the following facial movements: 1) frowning (frontal muscle), 2) eye closure (orbicularis oculi muscle) separately for B1 upper lid and B2 lower lid movement, 3) nose wrinkling (nasal muscles), 4) showing teeth (smiling; zygomatic muscles), 5) pursing lips (orbicularis oris muscle), 6) depressing lips (depressor anguli oris muscle), and 7) neck movement (platysma). The allocation of the visible movement to muscles was controlled by prior electric recording from the underlying mimic muscles, but continuous electric monitoring during surgery was performed only from the muscles A, B1, E, and F (see above). Each investigator documented independently the facial movements for each facial nerve branch in all cases. Finally, the results were compared, and all cases with divergent evaluation were reviewed together until a consensual assessment was obtained. A disagreement in the first evaluation between the two reviewers occurred in three out of the seven patients for 16 out of 612 movement analyses without discernable pattern; overall 68 facial nerve branches were stimulated and nine different movements evaluated.

Peripheral Branching of the Facial Nerve
During parotidectomy, all peripheral facial nerve branches were dissected systematically in an anterograde direction beginning from the main trunk. All branches were pursued until the anterior border of the parotid gland. Three examples of the peripheral facial nerve branching are shown in Figures 1 to 3. As branching of first order, the main trunk (0) always divided into a temporofacial division (branch 1) and a cervicofacial division (2). A trifurcation was not seen. In one case a nerve loop was seen between the main trunk and the temporofacial division (Fig. 2). This loop was clearly and reliably excitable by electric stimulation. As branching of second order, the temporofacial division divided either into two branches, an upper branch (11) and a lower  other four cases did not show a further subdivision until the anterior border of the parotid gland. The lower branch of the temporofacial division (12 and/or 13) never showed a further subdivision. The upper branch of the cervicofacial division (21) also showed no further subdivision. The lower branch of the cervicofacial division (22) divided into an upper (221) and lower branch (222) in two cases. The other five cases presented no further branching until the anterior border of the parotid gland.
In summary, all patients showed within the parotid gland a bifurcation of the main trunk into a temporofacial and a cervicofacial division. Furthermore, all patients showed a branching of the temporofacial division into at least two branches (second order branching). All but one patient also showed a branching of the cervicofacial division into at least two branches within the plane of the parotid gland. Branching of the third order was very variable and maximally seen in three patients within the plane of the parotid gland.

Electrostimulation of the Peripheral Facial Nerve
The results of the stimulation of the branches of the peripheral facial nerve are summarized in Table I. Stimulation of the main trunk of the facial nerve led to a reliable and visible stimulation of the whole hemiface in all but one patient. Neck movements were reliably visible only in four cases. Interestingly, the stimulation of the temporofacial division led to reliable eye closure but at the same time also to movements in the midface related to the function of the zygomatic muscles (in six out of seven patients). Stimulation of the cervicofacial division was followed by activation of the midface in only one patient but always of the mouth and mental region related to the orbicularis oris muscle and the depressor anguli oris muscle. Neither any branch of the temporofacial division nor of the cervicofacial division (of a second or third order) led to a movement related to single mimic muscle function.
From the perspective of the different mimic muscles, an allocation of specific peripheral facial nerve branches to a specific movement or muscle movement was not possible (compare Table I and Figs. 1-3). Stimulation of the whole spectrum of branches of the temporofacial division could lead to eye closure (orbicularis oculi muscle function). Stimulation of the spectrum of nerve branches of the cervicofacial division could lead to reactions in the midface (nasal and zygomatic muscles) as well as around the mouth (orbicularis oris and depressor anguli oris muscle function). A separate stimulation of upper versus lower eye lid was possible only in some patients. From a functional point of view, the frontal and eye region (frontalis muscle and orbicularis oris muscle) was exclusively supplied by the temporofacial division. The region of the mouth (orbicularis oris, the depressor anguli oris muscle area) and the neck were exclusively supplied by the cervicofacial division. There was a functional overlap in the midface. The nose and zygomatic region were mainly supplied by the temporofacial division, but two patients had also nerve branches of the cervicofacial division functionally supplying the nasal and zygomatic region.

DISCUSSION
The presented analysis showed that the topography of a facial nerve branch dissected during parotidectomy does not allow reliable conclusions on the related target muscle and facial movement. The innervation pattern of the peripheral branches, especially distal to the temporofacial and cervicofacial division was highly variable. Furthermore, within the parotid gland, an allocation of a nerve branch to an individual muscle function was not possible. Most stimulations led to synchronous movements of different mimic functions. Looking from the muscle side, what is important for the reliability of standard facial nerve monitoring, monitoring of an individual mimic muscle does not automatically mean that all peripheral facial nerve branches of the same region are monitored.
The methodology used had limitations. Only seven patients were investigated with a time-consuming electrostimulation mapping of the peripheral facial nerve. Nevertheless, the high variability of the anatomical facial nerve branching and the highly variable functional connection of the electrostimulated nerve branch and facial target region suggest that this high variability is a common phenomenon of peripheral facial nerve function. Only the facial nerve branches within the parotid region were analyzed. Therefore, we cannot exclude that more distal facial nerve branch stimulation might lead to selective stimulation of individual mimic muscles and facial movements. It seems to be unlikely that unintended antidromic stimulation along the facial nerve trunk caused some of the stimulation effects via an F wave. Much higher supramaximal stimulation is needed to trigger a facial nerve F wave. 11 A comparison to other studies is not possible, as comparable studies were not performed for the facial nerve. In contrast, the innervation of laryngeal muscles has been investigated in detail in the recent years by functional electrostimulation providing a better understanding of laryngeal neuroanatomy. [12][13][14] The two latter studies were performed intraoperatively in two patients, respectively. This should be taken into account when considering that the present study with high intraoperative effort of time was performed in only seven patients.
From many anatomical studies it is well known that the five main extratemporal branches of the facial nerve (frontal, zygomatic, buccal, marginal, and cervical) are anatomically connected not to a single target muscle but with some variation to several mimic muscles, and that these five peripheral branches exhibit, again with a high degree of variability, a variety of nerve fiber interconnections and sometimes also nerve loops within the parotid gland. 4,6,7,15 It has to be emphasized that even the terminal branches outside the parotid gland (distal to the facial plexus within the parotid gland) often show temporal-zygomatic, zygomatic-buccal, buccal-mandibular, and mandibular-cervical nerve anastomoses. 16 We can now confirm that this anatomic condition is mirrored by the physiological functional situation.
For facial nerve monitoring during parotid surgery, usually two or four channels are used for the recording from the facial muscles. 17 For the mostly used two-  channel monitoring, the needle electrodes are most frequently inserted in the superior portion of the orbicularis oris and into orbicularis oculi muscles. These recording sites are sufficient to monitor the main trunk of the facial nerve, but will lead to false-negative results if peripheral facial nerve branches that innervate other facial muscles are stimulated or injured. The present study showed that there are several constellations, especially for intraparotid nerve branches beyond the first branching, where two-channel or even four-channel monitoring might not be reliable. Facial nerve reanimation surgery and muscle transfers are the standard procedures for reconstruction of facial function. 18 Nevertheless, functional results are limited, and surgery is often not possible in cases of chronic denervation. Therefore, bionic devices with artificial electrical stimulation of extramuscular facial nerve branches, intramuscular nerve branches, or direct stimulation of facial muscles have been discussed for some time. Recent studies have shown the feasibility of facial pacing in animal models and in humans. 9,19,20 Successful rehabilitation via facial nerve branch pacing will depend on the selectivity of the electrical stimulation. For instance, for selective pacing of eye closure, it will be necessary to stimulate the orbicularis oculi muscle itself, using an electrode array with contact to large parts of the muscle. 21 Alternatively, one could stimulate a facial nerve branch, selectively innervating the orbicularis oculi muscle. The present results suggest that the way of extramuscular facial pacing will be successful when targeting peripheral facial nerve branches, sometimes even beyond the parotid gland.

CONCLUSION
This prospective intraoperative clinical study of seven patients demonstrated that a uniform functional allocation of specific peripheral facial nerve branches to a specific mimic movement was not possible. Functionally important facial movements, such as eye closure or movement for the opening of the mouth, were related to several peripheral nerve branches. Such a redundant innervation might explain why a lesion of a peripheral nerve branch within the parotid gland does not automatically lead to facial nerve dysfunction. It was frequently not possible to activate a specific isolated muscle function by intraparotid peripheral facial nerve stimulation, which has implications on planning the site of nerve stimulation of future bionic devices to restore facial nerve function. Furthermore, the present study showed that there are several constellations, especially for intraparotid nerve branches beyond the first branching, where two-channel or even four-channel monitoring might not be reliable.