Seeing Is Not Believing: Intraoperative Nerve Monitoring (IONM) in the Thyroid Surgery

Abstract

Ensuring the integrity of the recurrent laryngeal nerve (RLN), the external branch of superior laryngeal nerve (EBSLN) and preservation of normal voice are the prime ‘functional’ goals of thyroid surgery. More in-depth knowledge of neuronal mechanisms has revealed that anatomical integrity does not always translate into functional integrity. Despite meticulous dissection, neural injuries are not always predictable or visually evident. Intraoperative nerve monitoring (IONM) is designed to aid in nerve identification and early detection of functional impairment. With the evolution of technique, intermittent monitoring has given way to continuous-IONM. Over the years, IONM gathered both support and flak. Despite numerous randomised studies, systematic reviews, and meta-analyses, there still prevails a state of clinical equipoise concerning the utility of IONM and its cost-effectiveness. This article inspects the true usefulness of IONM, elaborates on the optimal way to practice it, and presents a critical literature review.

Introduction

The external branch of superior laryngeal nerve (EBSLN) and recurrent laryngeal nerve (RLN) have been the Achilles heel for thyroid surgeons since time. Amelita Galli Curci’s anecdotal tale, who lost her ability to sing at high pitch due to EBSLN injury, is well known and has sensitised surgeons to its potential morbidity. Depending on the severity of the neural injury, the presentation may vary from hoarseness of voice to feeding tube dependency and rarely tracheotomy due to bilateral vocal cord palsy. In the initial ages of thyroid surgery, the perception was never to unearth the nerves, but later Lahey popularised identifying the RLN [1]. We have progressed to an era where we not only visualise but also assess the nerve’s functionality during surgery.

This article analyses the applicability of IONM and the optimal practices and presents a critical review of the literature and prospects.

Utility of IONM

The application of IONM aids the surgeon in the early identification of the RLN and SLN. This improves the surgical performance by decreasing stress [2], especially in high-risk cases such as revision surgery, thyrotoxicosis, multiple central compartment lymphadenopathy, locally advanced thyroid cancer, and large goitre with retrosternal extension [3–12]. Its use significantly improves temporary and permanent RLN palsy rates in these high-risk cases [12–14], where visual inputs are not ample. It also helps localisation in rare anatomical variations like extra-laryngeal branching or non-recurrent nerve (<1%) [15–18].

Visual examination identifies only 11.3–14% of all injured nerves [19, 20]. In contrast, IONM is highly precise in predicting functionality because of its high (90–100%) negative predictive value [10, 21–24]. This advantage is attributed to identifying non-transection injuries like traction, thermal damage, compression, and clamping [24, 25]. Retrograde tracing may help localise the injury site so that any reversible damage like clipping can be undone. Continuous-IONM (C-IONM) reduces neuropraxia caused by traction injury and helps in real time to abort the causal manoeuvre [25–27].

It helps the neophyte surgeons to dissect out the RLN after initial identification at one point by creating an anatomical road map [28–30] and increasing the speed of detection (by 35%) [31]. This also enables the reduction of the volume of normal thyroid remnant left behind after resection [10, 18].

The advent of IONM has changed the way infiltrated nerves are managed intraoperatively. The degree of RLN invasion ranges from superficial abutment to complete deep infiltration. McCaffrey et al. reported 47% RLN invasion in papillary thyroid cancer [32]. The neural invasion may be divided into two types—type A, invasion limited to epineurium only, and type B, invasion beyond the epineurium fibrous layer with tumour extension to perineurium or endoneurium with abutment around the axons [11].

Infiltrated RLN may show a normal pre-operative laryngoscopy in 45% cases [33] and residual electrophysiologic activity on electromyography (EMG). This forms the basis of the management algorithm of type B RLN invasion (perineurium and endoneurium), which cannot be assessed on visual examination alone. The decision to resect or preserve the nerve is guided by various factors such as age, aggressive histology, the opposite cord status, distant metastasis, iodine avidity, and proximal EMG signals obtained using IONM [11, 33].

After the loss of signal (LOS) on one side, if surgery proceeds, bilateral vocal cord palsy incidence is 16.7–18.8% [34, 35]. Hence, surgery can be selectively staged for the contralateral side if there is no spontaneous recovery after 20-min wait period [5, 25, 36]. Staged surgery for the contralateral side has shown to decrease bilateral vocal cord palsy. The identification of EBSLN has also increased with IONM [37], which helps in its preservation [38].

General Setup and the Stepwise Checkpoints for Successful Use of IONM [25, 26]

Various invasive (needle electrodes) and non-invasive IONM methods are available to identify the nerves during thyroidectomy. However, endotracheal tube-based surface electrodes are the most commonly used and least variable method. Both intermittent and continuous stimulation methods are available, and a combination of electromyography (EMG) with an audio signal is considered superior to audio alone.

Parts of Monitoring Unit

The stimulation part includes the stimulator probe. The monopolar probe is better than bipolar. It provides a diffuse current spread and helps to search for a larger area. The bipolar probe is used for focal stimulation and has limited use. The reference electrode is placed more distally over the shoulder or sternum to minimise the stimulus artefacts. The stimulating-side ground electrode is placed at least 10 mm away from the reference electrode to minimise other devices’ artefacts.

The recording part involves the endotracheal tube with four pre-fashioned surface EMG recording electrodes (two for each vocalis muscle). A recording-side ground electrode is placed near the surgical site. Poor grounding may lead to noise artefacts.

Electro-neurophysiological (ENP) Unit

The ET tube electrodes, recording-side ground electrode, stimulator-side reference electrode, and stimulator probe are appropriately set up with the connector box. The interface connector box should be connected to the EMG monitor (Fig. 1).

Fig. 1.

Operating room layout for IONM. Abbreviations: EMG, electromyogram; ETT, endo-tracheal tube

Optimal IONM Checklist

Table 1 shows the best practices that are to be followed to avoid pitfalls [25, 26, 39].

Table 1.

Optimal IONM checklist [38]

Steps	Optimal checklist
Step 1	Pre-operative period
	Pre-operative laryngoscopy to assess the vocal cord mobility
Step 2	The anaesthesiologist’s checklist: before and during induction
	1. Appropriate size endotracheal tube (ET tube) with pre-fashioned surface EMG electrodes to optimise the contact of electrodes with the vocal cords and ensure low impedance
	2. Lubricant or lidocaine jelly on the ET tube is avoided
	3. Use short-acting neuromuscular agents for induction
	4. After giving the final surgical position to the patient, the tube position is confirmed to mark 12 o’clock by direct vision or video laryngoscopy
	5. Medication and mechanical suctioning are used to decrease the pooling of saliva near the recording electrodes if required
	6. After intubation, neuromuscular blockers are not repeated as they reduce EMG amplitude and decrease the sensitivity of neural signals
	7. Proper depth of anaesthesia is maintained to prevent spontaneous vocal cord activity
Step 3	The neurophysiologist’s checklist: electro-neurophysiological (ENP) unit
	The ET tube electrodes, recording-side ground electrode, stimulator-side reference electrode, and stimulator probe are set up with the connector box appropriately. The interface connector box should be connected to the EMG monitor
	1. By tapping the finger on either side of the thyroid cartilage (Tap test), completeness of the circuit is confirmed before incision
	2. Electrocautery unit should be placed more than 10 ft away from the neural monitoring unit. Muting cables are used to inactivate the EMG monitor temporarily during cautery use
	3. It does not get affected by cardiac pacemaker or vessel sealing energy devices
	4. Setting for the IONM
	a. A low impedance of <5 K Ohm suggests good electrode patient contact and impedance imbalance of <1 K Ohm between two sides is allowed
	b. Stimulation probe current is kept at 1–2 mA with a pulsatile output of 4 Hz/s with 100 μs pulse duration
	c. The threshold of the monitor event is set at 100 microvolts (μV). A too high threshold may not detect the signal
	b. Human RLN depolarises maximally at 0.8 mA; hence a safe suprathreshold 2 mA current is used for searching and mapping; later on, a lower 0.3–0.5 mA is used for assessing functional integrity
	e. The amplitude of the waveform will be between 100 and 1500 μV. In general, an amplitude of >250 μV is highly suggestive of intact functioning nerve. The latency of the left side will be longer than the right side due to anatomical dissimilarity on vagus stimulation
Step 4	The surgeon’s checklist: during surgery
	1. The strap muscle test is performed to confirm the functioning of the stimulator probe by observed muscle twitching. Stimulator probe is best used by dragging over the tissue as it results in better tissue contact in between the pulses
	2. V1: Vagus nerve stimulation is done by a supra-threshold current before starting the dissection. This confirms the functionality of the circuit and helps in detecting the true negative signals. For V1, vagus is stimulated at a level above the thyroid cartilage and at a caudal point
	3. R1: RLN is searched for and identified by stimulation and amplitude of EMG waveform is recorded
	4. R2: RLN to be stimulated again at the end of the procedure
	5. V2: Suprathreshold vagal stimulation is done again at the end of the procedure. This helps to detect the false-negative incidences due to stimulation of RLN distal to the site of injury during R2
	6. Compare and document the EMG signals’ amplitude before (V1, R1) and after the dissection (R2, V2). Even a 50% decrease in amplitude or a 10% increase in latency suggests a damaged nerve
Step 5	Post-operative period
	The vocal cord function assessment is done within/at 2 weeks postoperatively even in the absence of symptoms

Non-recurrent Laryngeal Nerve (NRLN) [16, 25, 40, 41]

NRLN is a rare entity with the incidence reported in a large meta-analysis by Henry et al. as 0.7% on right (53 studies, n = 81/33,571) and that of left NRLN being <0.01% (41 studies, n = 2/20,006) [40]. Toniato et al. has reported a higher intraoperative injury rate (12.9%) than that reported for the normal anatomic course. They further proposed the following anatomic classification [41]:

• i)Type 1: It arises high from the cervical vagus nerve and descends along with the superior thyroid vessels.
• ii)Type 2A: It rises from the cervical vagus nerve lower to the laryngotracheal junction level. It follows a transverse path parallel to the inferior thyroid artery.
• iii)Type 2B: It rises from the cervical vagus nerve lower to the laryngotracheal junction level and drops below the inferior thyroid artery either dorsal to it or in between its branches and then travels cranially ascending to the laryngeal entry.

The presence of NRLN can be ascertained by:

1. V1 stimulation: Presence of signal from vagus at a higher point (above superior cornu of the thyroid cartilage) and absence at a caudal point (below 4th tracheal ring) suggests NRLN (Fig. 2).

2. Lower latency (<3.5 ms) has also been reported due to the shorter course of NRLN.

Fig. 2.

A V1 stimulation: to detect non-recurrent laryngeal nerve before dissection. Abbreviations: RLN, recurrent laryngeal nerve; NRLN, non-recurrent laryngeal nerve; I, IIA, and IIB: types of non-recurrent laryngeal nerve; ITA, inferior thyroid artery; Vagus N, vagus nerve. B Method of assessing the laryngeal twitch

NRLN is less common on the left side, and it is associated with dextrocardia. Therefore, the above manoeuvre is done only on the right side unless there is dextrocardia. Abnormal anatomy of the RLN is detected before starting the dissection, alerting the surgeon about the variation.

Superior Laryngeal Nerve Monitoring [37–44]

EBSLN is the unsung nerve during thyroid surgery. However, injury leads to loss of high-pitched voice, abnormal mucosal waveform, bowing and lowering of the vocal fold, and ipsilateral laryngeal rotation.

The monopolar probe is used, and the area near the laryngeal head of the sternothyroid muscle is stimulated. Stimulation of EBSLN causes cricothyroid muscle twitch, which is visualised. A needle electrode or specially designed ET tube with additional electrodes may be placed to get the EMG data recording.

Algorithm for Identifying the Causes of Loss of Signal (LOS) [11, 25, 26, 39]

True LOS is defined as the complete absence of signal when stimulated with a current of 1–2 mA in dry field or absence of laryngeal twitch on ipsilateral vagal stimulation, and at the same time, there was an EMG activity of amplitude >100 μV, to begin with.

The laryngeal twitch is palpated with a finger behind the posterior lamina of the cricoid cartilage during ipsilateral vagal stimulation. The presence of twitch indicates the contraction of posterior cricoarytenoid muscle and functioning RLN (Fig. 2B).

Refer to Fig. 3 to follow the algorithm for LOS [25, 26, 39].

Fig. 3.

Management algorithm for loss of signal (LOS). Abbreviations: RLN, recurrent laryngeal nerve

As defined by the International Neural Monitoring Study Group Guideline, the following are the types of RLN injuries depending on the LOS [25, 45, 46]:

• Type 1 RLN injury: Clear-cut segmental RLN injury, mainly caused by direct trauma to RLN.

• The most frequent is between the intersection of RLN with the inferior thyroid artery (ITA) and ligament of Berry, followed by at the intersection of the RLN with ITA, and, lastly, below the intersection of the RLN with ITA.

• Type 2 RLN injury: The whole of RLN is nonconductive. This implies more global or diffuse damage and maybe with an intralaryngeal focus.

• Type 1 RLN injuries are more severe than type 2 RLN injury [46]. Hence, type 2 has a better clinical outcome than type 1; also, there is a possibility of corrective action if LOS is detected early during dissection.

False-negative:

• Injury after the last testing—e.g. during wound irrigation or haemostasis.

• Stimulating distally to injured RLN segment without performing V2 at the end.

• Posterior RLN branch injury: Posterior cricoarytenoid muscle is not assessed by ET-based IONM system. Additional post-cricoid surface electrodes are required for the same.

• Non-neural issues: Laryngeal oedema, arytenoid cartilage dislocation.

• Progressive oedema leading to delayed neuropraxia at an intralaryngeal site or a delayed vascular cause.

False-positive:

• Neuromuscular blocking drugs

• Endotracheal tube displacement

• Blood or fascia covering the nerve

• Early responses due to the stimulation suppression artefact

• Early neural recovery after transient neuropraxia lasting for a shorter time

Current Evidence—Issues with Interpreting Existing Literature on IONM and Cost-Effectiveness of IONM

The evolution of the technology on IONM for thyroid surgeries has generated vast amounts of literature over the last few decades. Despite numerous systematic reviews and meta-analyses (Table 2) [8–10, 12, 47–56], there still prevails a state of clinical equipoise concerning the utility of IONM, especially in preventing permanent vocal cord palsy. Critical appraisal reveals the following issues associated with the interpretation of the available literature; however, none of the guidelines recommended against the use of IONM.

1. Literature is comprised mainly of retrospective case series (non-comparative studies) and few randomised studies. Though the cumulative number of patients is large with increased statistical power, the data is still heterogeneous.

2. More advanced cases were allotted for IONM, resulting in definite selection bias.

3. Foreign language studies were not included in meta-analyses. None of the series collectively studied the quality of life and other dependent parameters (e.g. hypoparathyroidism, local control rate, mortality, residual Radioiodine uptake) [56].

4. The randomised trials included in the meta-analyses have a smaller sample size and a low methodological quality [2, 21, 57–59]. Also, the trials were not adequately powered to establish the superiority of IONM. Since the incidence of RLN palsy is low (temporary rate 6–9.8% [25, 26, 60, 61] and permanent 3% [62]), the sample size required to demonstrate the superiority of IONM has been postulated to vary from 7000 to as high as 40,000 [5, 21, 49], which is indeed very challenging.

5. High-risk cases (malignancy, revision, large goitre, Graves’ disease, retrosternal extension, higher nodal burden) have shown benefit with IONM; however, what constitutes “high risk” has not been defined [5, 9, 10, 12, 63]. There is a lack of well-powered focusing on this issue. Lack of adequate pre-operative knowledge about the degree of surgical difficulty (risk stratification) is another concern.

6. The extent of surgery performed across studies was variable-total thyroidectomy, subtotal thyroidectomy, lobectomy, and revision surgeries (information regarding the exploration of the earlier operated side is inadequate/incomplete) [5, 58, 64]. Also, different histologies were pooled together, adding to the heterogeneity.

7. Intermittent-IONM (I-IONM) helps trace the nerve and prognosticate functionality when the damage has happened. But the event of temporary RLN injury as such is not prevented [65] as the damage to the nerve can occur in between two stimulations while using intermittent IONM. The use of continuous IONM (C-IONM) technique can overcome this problem. It is a more promising technique; however, it is not routinely used [27]. Sensitivity for prediction of VCP by I-IONM is 63–91.3% and 90.9–100% for C-IONM. Specificity for VCP prediction is 97.1–99.5% for I-IONM and 90.2–99.7% for C-IONM [66].

8. Though identification is improved by IONM, handling the nerve solely depends on surgical skill, experience, operative volume, and surgical difficulty level (RLN relation to thyroid gland/neoplasm and scar tissue). Despite the fact that these salient factors vary from surgeon to surgeon and case to case with different outcomes, all data have been clubbed together to analyse the efficacy of IONM [63, 67–70].

9. There is non-uniform assimilation of essential data such as pre-operative and post-operative laryngoscopy for vocal cord status across various studies (done only in 50.1% cases), thereby affecting the true predictive value of IONM [19, 25, 26, 71, 72].

10. The technique of laryngoscopy and timings of the assessment (early vs until 2 weeks) vary across studies. Also, the permanent vocal cord palsy criteria were different across the reviews (6 months vs 12 months).

11. The permanent vocal cord palsy rate is much lower than the temporary vocal cord palsy rate. Hence to prove its usefulness, it will require a very high sample size.

12. Since the technology of IONM has evolved over the decades, the heterogeneity across studies may also impact the results of various meta-analyses and systematic reviews.

13. Studies have indicated that the initial years of utilising IONM are associated with higher technical issues and inadvertent RLN injuries, which decrease over time as the surgeons and the centre go through the natural learning curve and familiarising themselves with the technique. The routine use of IONM may be helpful to shorten the learning curve faster [64, 65, 73, 74].

14. The absence of an objective and uniform detection method has likely led to underreporting of the EBSLN palsy even in randomised trials focusing on the EBSLN injury [44, 75–79].

Table 2.

Summary of meta-analyses and systematic reviews.

A well-conducted randomised study with a homogenous population may dissolve this debate. We have an ongoing prospective randomised study to address the same controversy (ACTION study; PI: Dr. G Pantavaidya; CTRI/2014/09/007644).

Is IONM Cost-Effective?

Literature is equivocal on the issue of cost-effectiveness of IONM with studies both for and against [80–84]. This is likely due to various methods of cost-effectiveness analysis utilised and variations in assumptions in each model. Al-Qurayshi et al. used the Markov model and postulated that IONM is more economical than routine visual inspection [80, 81]. Simulation economic modeling by Wong et al. showed the cost-effectiveness of IONM in preventing permanent RLN palsy [82]. Rocke et al., in their analysis using a decision tree model, concluded contrary to the above, mentioning that IONM would only be cost-effective if the surgeon was able to reduce his nerve palsy rate by more than 50.4% using IONM as compared to visual inspection, implying an unrealistic high palsy rate [83]. Sanabria et al. also used a similar decision tree model and found that nerve palsy rates were comparable with and without IONM (1% vs 1.6%) [84]. Moreover, cost-effectiveness is also governed by factors such as the kind of healthcare system (universal/non-universal), the average cost of surgery, and patient preferences, which vary across the geography and socio-economic status of a country. The direct and indirect cost of rehabilitation, legal claims, phonosurgery, and time to recover—all are important and need to be accounted for [85–87].

C-IONM-Potential Paradigm Change

Continuous IONM is performed by the ipsilateral vagus stimulation by a specially designed stimulating electrode, and the nerve is stimulated continuously by automatic periodic stimulation (APS). This APS is performed by delivering 1 mA current at the stimulation rate of 1 Hz with the pulse width of 100 μs continuously [88]. The ET tube electrode is used, and latency and amplitude are captured. The minimum baseline amplitude would be established with 500 μV.

This improvement in technology has the potential to avoid neural injury, which may be caused in between two intermittent RLN stimulations while using the traditional I-IONM. It is real-time and abnormal signal transmission alerts the surgeon before the irreversible damage [25, 26, 89–91]. Continuous stimulation of the vagus has raised concerns regarding hemodynamic instability [92]. However, many studies have proven its safety in an atrioventricular block and paediatric population without any significant cardiopulmonary, gastrointestinal, or pulmonary adverse effects [93–95].

Negative Attributes of IONM

1. Although IONM is associated with a high negative predictive value, it is known to yield a low positive predictive value (12–88%) [21–26], resulting in unnecessary delay or aborting procedure for a second stage [96, 97].

2. The learning curve of IONM is long and associated with inadvertent neural damage in the initial period due to a false sense of security and inexperience with the technology [65].

3. IONM requires additional technical staff to handle the nerve monitoring unit and adds to cost [80–84].

4. As compared to the traditional technique, the use of IONM may increase the operative time. However, few studies claim to reduce the duration by early nerve identification [31].

5. IONM adds to the cost incurred to the patient, and with inadequate evidence to support its routine use, advocating the technology for all patients may not be appropriate. Similarly, backing IONM citing medicolegal safety without conclusive evidence may also not be fair, though it has been made mandatory in select countries [85–87].

Conclusion

The evolution of IONM has brought about a paradigm shift in how we handle and preserve the nerves during thyroid surgery. The technological advancement from intermittent to continuous IONM use, standardisation of procedure for better quality, and uniform guidelines can significantly reduce vocal cord palsy rate and guide intraoperative decision-making. While there is a lack of consensus regarding its benefit in routine cases, its high negative predictive value has shown a definite advantage in high-risk procedures. It is an educational surgical adjunct, and the cost of the technology needs to be balanced carefully against its putative benefit, especially for medicolegal issues.

Data Availability

Not applicable.

Declarations

Ethics Approval

Not applicable.

Consent to Participate

Not applicable.

Consent for Publication

Attached separately.

Conflict of Interest

None of the authors have any conflict of interest with respect to this manuscript.