Efficacy of intraoperative recurrent laryngeal neuromonitoring during surgery for esophageal cancer

Abstract Aim To evaluate the efficacy of intraoperative neuromonitoring in identifying recurrent laryngeal nerves and decreasing the incidence of nerve injury in minimally invasive esophagectomies for esophageal cancers. Methods A total of 167 minimally invasive esophagectomy patients were retrospectively reviewed. They were divided into intraoperative neuromonitoring (n = 84) and no intraoperative neuromonitoring (n = 83) groups, based on whether or not intraoperative neuromonitoring was used during surgery. We compared short‐term surgical outcomes and incidence of recurrent laryngeal nerve palsy between the two groups before and after propensity score matching. The association between the loss of signal and recurrent laryngeal nerve palsy was also evaluated. Results The incidence of recurrent laryngeal nerve palsy (grade 2 and higher) was lower in the intraoperative neuromonitoring group than in the no intraoperative neuromonitoring group (6.0% vs 21.2%, P = 0.02). The rate of recurrent laryngeal nerve palsy recovery within 6 months was also significantly higher in the intraoperative neuromonitoring group (87.5% vs 20.0%, P < 0.01). The positive and negative predictive values of intraoperative neuromonitoring for recurrent laryngeal nerve palsy were 60% (9/15) and 86.9% (60/69), respectively. The duration from paralysis to recovery was shorter in recurrent laryngeal nerve palsy cases with negative loss of signal results than in cases with positive loss of signal results (median: 43 days vs 95 days). Conclusion Intraoperative neuromonitoring is useful in identifying recurrent laryngeal nerves and may aid in reducing the incidence of recurrent laryngeal nerve injury during esophageal surgery.


| INTRODUC TI ON
Esophageal cancer is associated with a high incidence of lymph node metastasis and often requires the use of esophagectomy with radical lymphadenectomy. 1 In recent years, owing to the development of endoscopic equipment and progress in surgical techniques, minimally invasive esophagectomy (MIE) has become the treatment of choice in patients with esophageal cancer with locoregional lymph node metastasis. The advantages of MIE over conventional open transthoracic esophagectomy have been reported in several studies. [2][3][4][5] However, an increased frequency of postoperative complications such as recurrent laryngeal nerve palsy (RLNP) has been observed; therefore, care must be taken during surgery. 6 Although delicate surgical procedures can now be performed owing to the enlarged view offered by endoscopy, the recurrent nerve is vulnerable to injury due to thermal damage or towing, and this could cause postoperative vocal cord palsy and swallowing dysfunction.
Furthermore, as postoperative complications affect patients' longterm prognoses, 7 preventing RLNP is very important.
Intraoperative neuromonitoring (IONM) has gained widespread acceptance as a useful tool for the preservation of nerve function during thyroid surgery and recently during surgery for esophageal.
Few studies have focused on the importance of IONM in surgeries for esophageal cancer, and its usefulness has not been revealed.
Therefore, in this study, we aim to evaluate the efficacy of IONM in identifying RLN injury and reducing the incidence of RLNP.

| Data collection
Preoperative tumor stage evaluations included: a medical interview; physical examination; upper gastrointestinal endoscopy and biopsy; computed tomography scan of the chest, abdomen, and pelvis; positron emission tomography-computed tomography; and blood tests.
Initial evaluation revealed that none of the patients had RLN paresis or paralysis before surgery.
Clinical data of all patients were collected from a database at our institution. Tumor staging was done according to the UICC esophageal cancer TNM staging system (8th edition). For postoperative complications, RLNP was divided into grades 1 and 2 or higher using the Clavien-Dindo (CD) classification. RLNP requiring the use of antibiotics for the treatment of aspiration pneumonia was classified as grade 2. Aspiration pneumonia was defined as follows: (a) pneumonia following a witnessed macroaspiration event, in which case the content of aspiration was confirmed in the trachea; (b) a swallowing video endoscopy-confirmed aspiration and repeated symptoms of aspiration; (c) a computed tomography scan showing infiltrative findings in the superior or basal segment of the lower lobes, or the posterior segment of the upper lobes of the lungs.

| Surgical procedure
We performed three-field esophagectomy with an anastomosis in the neck and started with the thoracic component. Right thoracoscopic access was secured, and four trocars were inserted. A transitory CO 2 pneumothorax (6 mm Hg) was instituted in the prone position, followed by esophageal mobilization and mediastinal lymphadenectomy. The lymph nodes around the RLN were dissected using an ultrasonic scalpel and cold cutting scissors ( Figure 1).
A gastric conduit was constructed and perigastric lymphadenectomy was performed in the supine position along with hand-assisted laparoscopic surgery. Through a collar-shaped neck incision, both RLNs were identified before the removal of the cervical paraesophageal nodes. For patients with mediastinal and perigastric lymphadenectomy as D2 dissection, right cervical paraoesophageal node dissection was completed during the transthoracic procedure. In D3 lymph node dissection, both supraclavicular lymphadenectomy was performed through the neck incision in addition to D2 dissection. The gastric conduit was pulled through the posterior mediastinum or retrosternal routes, and cervical esophagogastric anastomosis was performed using an end-to-side circular stapler.
Three surgeons performed the surgeries. The efficiency of each surgeon was classified based on the number of procedures they had performed (either ≥ 31 or ≤30).

| IONM
In IONM, the evoked electromyogram (EMG) of a muscle is monitored by direct electrical stimulation of its innervating nerve with a nerve stimulation probe during surgery after an electrode is set on the muscle. Neural function was confirmed by the sound converted from the EMG response.
The patients in the IONM group were intubated under general anesthesia using the NIM TriVantageTM EMG tube (Medtronic ENT) and underwent intraoperative RLN monitoring by the NIM-response® system (Medtronic ENT) for EMG. Neural stimulation was done with a hand-held monopolar stimulator (Medtronic ENT), usually with currents of 1.0 mA. When stimulation of the RLN and the vagus nerve produced no EMG activity (electrical silence) or substantially reduced EMG activity (<100 mV), loss of signal (LOS) was confirmed, and neural injury was suspected.
In the intrathoracic procedure, before the RLN was fully exposed and visualized, the IONM probe was used to stimulate the cords in order to search for and identify RLN ( Figure 1A,D). EMG activity was observed in response to intermittent nerve stimulation at RLN branched off the vagus nerve at the right subclavian artery and at the aortic arch ( Figure 1C,F). If LOS was detected, distal points were stimulated for the identification of the site of nerve conduction impairment. Lastly, before the end of surgery, the cervical vagus nerve was stimulated to observe neural activity.
We reviewed the intraoperative video files of all patients with RLNP in order to identify the likely site of damage and verify the procedure that caused it.

| Examination of vocal cord movement
RLNP was confirmed seven days after the surgery using indirect laryngoscopy. An otolaryngologist observed the movements of patients' vocal cords to check for paralysis. RLNP was defined as fixation and disturbance of vocal cord mobility and the affected side was noted. Patients with RLNP were followed-up and re-examined using laryngoscopy every 1-3 months until the palsy improved. Recovery from RLNP was defined as complete improvement in the mobility of the affected vocal cord, after which follow-up with the otolaryngologist was deemed complete. Vocal cord palsy failing to resolve within 12 months was considered to be a permanent RLNP.

| Postoperative morbidity
There was no difference in the overall RLNP incidence rate be- *P-value for two groups compared using the Fisher's exact test; significance was set at a value lower than 0.05.
Univariate and multivariate analysis of the risk factors for RLNP with CD grade 2 or higher are shown in Table 5. Multivariate analysis also showed that IONM was independently and significantly related to RLNP with CD grade 2 or higher (odds ratio = 0.21,

| Propensity score matching
Propensity score matching was used to balance the clinical characteristics and potential confounders between these two groups.
After propensity score matching, the demographic and clinical characteristics before treatment were adequately balanced between the 47 pairs in the IONM and no IONM groups: standardized difference < 0.100 (Table 1). There were significant differences between the patients in the no IONM and IONM groups in terms of blood loss (280 mL vs 180 mL, P < 0.01) and postoperative fasting period (8 days vs 8 days, P = 0.02) ( Table 2). The number of CD grade 2 or higher RLNP cases were lower in the IONM group than in the no IONM group (6.0% vs 21.2%, P = 0.02) (  Table 6 shows the correlation between LOS and postoperative RLNP in all patients with IONM. The sensitivity and specificity of the absence of LOS on IONM for intact RLN was 90.1% (60/66) and 50% (9/18), respectively. The positive predictive value (PPV) was 60% (9/15). The false-response rate (i.e. LOS absence with RLNP) was 13% (9/69). The false-no response rate (i.e. LOS without RLNP) was 9.1% (6/60). Five cases of LOS during the cervical procedure and one case during thoracic procedure were observed.

| Evaluation of LOS in IONM
The video review identified the site of RLN injury. The causes of injury in PPV were burns with an ultrasonic scalpel in two cases, gripping with forceps in one case, dissection with scissors in one case, over traction in three cases, and cervical vagus nerve taping in two cases.
The Kaplan-Meier curve of recovery from RLNP after esophagectomy in patients with IONM is shown in Figure 2. The time from RLNP to recovery was shorter without LOS than with LOS (P = 0.029). The median durations of recovery from RLNP were 43 and 95 days in the IONM and no IONM groups, respectively.  no severe paralysis of grade 3 or higher was observed. The duration from paralysis to recovery was shorter in the false-response cases than in true LOS cases. Seddon reported that peripheral nerve injury can be classified into neurotmesis, axonotmesis, and neurapraxia. 28 Neurapraxia is unaccompanied by peripheral degeneration and recovers rapidly and completely. In false-response cases, RLNP may have recovered early after a transient, delayed conduction disorder. The mechanism by which paralysis occurs could not be specified, but the duration to recovery from RLNP was thought to be short because of the minor degree of nerve damage.

| D ISCUSS I ON
It seems that the identification and verification of RLN using IONM could reduce the incidence of axonotmesis and neurotmesis during surgery. However, in intermittent IONM, it is necessary to confirm the stimulus response frequently in order to confirm functional preservation of the RLN. In the no IONM group, there was one case of intraoperative direct nerve injury, and in many cases, it was difficult to identify the site of injury. The low recovery rate in this group may be because the interruption of the axon or perineural nerves which led to the nerve palsy, did not recover due to misdirected innervation.
Recently, methods for the quantification of amplitude and latency for continuous monitoring have been reported, and these may be able to identify nerve damage more realistically and prevent nerve damage than intermittent methods.
Our findings show that intermittent IONM in MIE is easy to use and reduces the incidence of RLNP. This study has some limitations.
It had a single-center, retrospective design with a small sample size.
A randomized controlled trial should be performed in such settings, as the details of minor surgical procedures change over time. In our study, grade 1 mild paralysis could not be prevented, and false-response cases were observed. The newer real-time continuously neural monitoring devices may be more useful for preventing nerve palsy.
In conclusion, IONM was useful in the identification and verification of recurrent nerves and could reduce the incidence of RLNP.
We consider that IONM may increase the safety of esophagectomy and improve surgical outcomes.