Differential recurrent laryngeal nerve palsy rates after thyroidectomy Jonathan W. Serpell, MBBS, MD, MEd, FRACS, FACS,a James C. Lee, MBBS, FRACS, PhD,a Meei J. Yeung, MBBS, BMedSci, FRACS,a Simon Grodski, MBBS, FRACS,a William Johnson, MBBS, MD, FRACS, FRCS,a and Michael Bailey, PhD, MSc, BSc,b Melbourne, Victoria, Australia
Introduction. Recurrent laryngeal nerve (RLN) palsy is a devastating complication of thyroidectomy. Although neurapraxia is thought to be the most common cause, the underlying mechanisms are poorly understood. The objectives of this study were to examine the differential palsy rates between the left and right RLNs, and the role of intraoperative nerve swelling as a risk factor of postoperative palsy. Methods. Thyroidectomy data were collected, including demographics, change in RLN diameter, and RLN electromyographic (EMG) reading. Left and right RLNs, as well as bilateral and unilateral subgroup analyses were performed. Results. A total of 5,334 RLNs were at risk in 3,408 thyroidectomies in this study. The overall RLN palsy rate was 1.5%, greater on the right side than the left for bilateral cases (P = .025), and greater on the left side than the right for unilateral cases (P = .007). In a subgroup of 519 RLNs, the diameter and EMG amplitude were measured. The RLN diameter increased by approximately 1.5-fold (P < .001), and corresponded to increased EMG amplitude (P = .01) during the procedure. The diameter of the right RLN was larger than the left RLN, both at the beginning and end of the dissection (P = .001). Conclusion. The right-left differential rates of post-thyroidectomy RLN palsy seemed to be due in part to differential RLN diameters, with stretch having a more deleterious effect on RLNs with a smaller diameter; also, edema as a result of stretch might be an underlying mechanism for postoperative neurapraxia and palsy. Thyroid surgeons should be aware of the different vulnerabilities of each RLN and develop practices to avoid iatrogenic injury. (Surgery 2014;156:1157-66.) From the Monash University Endocrine Surgery Unit and Alfred Hospital,a and Department of Epidemiology and Preventative Medicine,b Monash University, Melbourne, Victoria, Australia
THYROIDECTOMY IS A COMMON OPERATION and one of its most devastating complications is recurrent laryngeal nerve (RLN) palsy.1-3 In a recent systematic review, the mean rates of temporary and permanent RLN palsy were 9.8% (range 1.4–38.4%), and 2.3% (range 0–18.6%), respectively.4 Functional consequences of RLN palsy range from dysphonia, weakened cough, and predisposition to aspiration, to life-threatening airway obstruction. A key aim of thyroid surgery is to avoid RLN palsy; to achieve this with a precise surgical technique, the thyroid surgeon needs a detailed Authors are listed in descending order of contribution. The authors thank their database manager Melissa Vereker for the extraction of the data from the database. Accepted for publication July 17, 2014. Reprint requests: Jonathan W. Serpell, MBBS, MD, MEd, FRACS, FACS, The Alfred Hospital, Commercial Road, Prahran, Victoria 3181, Australia. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2014.07.018
understanding of the surgical anatomy and neurophysiology of the RLN and to be aware of the clinical associations and the pathophysiologic causes of this complication.5 Anatomically, the left and right RLNs are different. The left is longer, usually 12 cm from where it winds around the ligamentum arteriosum to the larynx, whereas the right RLN is usually 6 cm from where it winds around the subclavian artery to the larynx. The left RLN tends to run within the tracheoesophageal groove, whereas the right RLN tends to run a more anterior and oblique course to the tracheoesophageal groove.6 Pathophysiologically, RLN palsy can be caused by stretching, heat, compression, devascularization, or transection, which results in neurapraxia, axonotmesis, or neurotmesis.7 Neurapraxia appears to be the most common mode of injury because most RLN injuries are not recognized at the time of operation with the nerves remaining visually intact. Most RLN palsies are temporary and resolve quickly.8,9 Electromyography (EMG) is a useful tool to assess nerve integrity intraoperatively because injuries SURGERY 1157
1158 Serpell et al
other than nerve transection may not be recognizable visually. Although the underlying cause of the neurapraxia is unclear, it has been reported in animal models that excessive tension can cause ischemia and myelin damage.10,11 It has also been reported that RLN diameter increases during thyroidectomy, presumably because of edema, and may be related to the subsequent neurapraxia.12 Therefore, we postulate that tension from intraoperative stretching can lead to decreased blood flow, and the subsequent edema perpetuates the ischemia and causes RLN neurapraxia in neck surgery. Asymmetric RLN anatomy modulating the tension from retraction was postulated as the reason for different rates of palsy between the left and right RLN after anterior cervical laminectomy.13,14 In the large number of thyroidectomy series reporting RLN palsy rates, only 2 studies in the past 5 decades have included the left-right breakdown; they all reported higher palsy rates on the right.15,16 At present, the underlying reasons for the discrepant RLN palsy rates between the left and right sides are poorly understood. There are insufficient data to support the role of the anatomic asymmetry in RLNs to explain the differential rates of right- versus left-sided RLN palsy. Therefore, the aims of this study were to: 1. Compare the left- and right-sided RLN palsy rates after thyroidectomy in a single surgical unit; 2. Investigate the correlation of RLN diameters to the differential palsy rates; and 3. Investigate the interaction between intraoperative RLN swelling and EMG changes, and their role in postoperative RLN palsy.
PATIENTS AND METHODS This retrospective cohort study was approved by the Human Research and Ethics Committee of The Alfred Hospital, Melbourne, Australia. Cases of thyroid procedures between 2007 and 2013 were recruited from the Monash University Endocrine Surgery Unit database. The information from this database was prospectively collected and audited regularly by the database manager. The patients for the measurements of RLN diameter and EMG were recruited between 2009 and 2011. Of the eligible cases, 6 were excluded from the study because of tumor invasion leading to intentional RLN sacrifice (4 cases) or incomplete documentation of the essential parameters (2 cases). Data were collected regarding the case demographics, RLN function pre- and post-thyroidectomy, change
Surgery November 2014
in RLN diameter during dissection, and functional status of the RLN according to the intraoperative neuromonitor (IONM) EMG reading. All procedures were performed under general anaesthesia with 4 mg of intravenous dexamethasone administered routinely on induction. All procedures were performed by 1 of 4 surgeons in the unit adhering to a standard method described previously.17 The RLN was first identified approximately where it was crossed by the inferior thyroid artery. It was then exposed widely by ligation of the tertiary branches of the inferior thyroid artery, resulting in gradual anteromedial rotation of the thyroid lobe.1-3,17 All surgeons used vessel-sealing devices; no thermal devices were used within 1 cm of the RLN. Hemostasis in the vicinity of the RLN was achieved by liga-clips, ligatures, or sutures. Results were analyzed as a unit; therefore, differential results between surgeons or thermal devices were not available. Bilateral procedures were defined as procedures involving both thyroid lobes in the one operative episode, therefore putting both RLNs at risk, and consisted mainly of total thyroidectomy. Bilateral reoperative procedures (eg, bilateral recurrence after previous bilateral subtotal thyroidectomy) were also included as bilateral procedures. Unilateral procedures mainly consisted of hemithyroidectomy (or lobectomy), but also included unilateral reoperative procedures. Sequential lobectomies, such as a diagnostic hemithyroidectomy followed by completion thyroidectomy in a subsequent operative episode were considered as two unilateral procedures, if both were captured during the study period. On initial identification, the diameter of the RLN was measured with vernier calipers with a resolution of 0.1 mm. The measurement was repeated just before the lobe was detached from the trachea at Berry’s ligament. The second measurement was made at the same point of the nerve where the first measurement was taken. The distance from inferior border of cricopharyngeus muscle was used as a reference to the position of the first measurement. If branching of the RLN was present, the diameter of the main trunk was measured. The IONM was performed with the NIM Nerve Monitoring System (Medtronics, MN), using the accompanying handheld neurostimulatory probe and the EMG endotracheal tube. The stimulation of the probe was set at 1.0 mA, which resulted in supramaximal stimulation of the RLN. The vagus nerve was stimulated routinely at the beginning and end of the case. If a muscle relaxant was used
Surgery Volume 156, Number 5
on induction of general anesthesia, either the muscle relaxant was reversed or the return of muscular activity was confirmed before EMG recording. Vocal cord adduction was recorded by the EMG endotracheal tube, and the threshold sensitivity was set at 100 mV.4,18 Abduction of the vocal cord was assessed by the laryngeal twitch technique, with palpation of contraction by the posterior cricoarytenoid muscle.5,18 The maximum amplitude achieved was recorded in the database. All patients had pre- and postoperative flexible fiberoptic laryngoscopy by a surgeon independent of the treating surgeon to assess vocal cord function. The postoperative assessment was performed one day after the operation but earlier if there were concerns regarding airway patency in the recovery room. Vocal cord assessment at this time point was the routine of the unit to maximize capture of temporary palsies and to minimize noncompliance after discharge. Subsequent vocal cord assessments were performed if an abnormality was detected at the initial postoperative assessment. The patient was monitored until resolution of the abnormality or definitive management of the palsy was instituted. A RLN palsy was regarded as permanent if it persisted over 6 months. Statistical analysis. Because some patients were included more than once if they had repeat operations, cases, procedures, or number of RLNs are referred to in this manuscript, instead of number of patients. All data were assessed initially for normality. Incidence rates were compared using the chi-squared test for equal proportions. Changes in RLN diameter were assessed by paired t tests and reported as means and standard errors (SE). All mean values are reported with SE values, unless stated otherwise. To account for repeated measures, the relationships between increase in RLN diameter and other parameters were assessed by generalized linear modelling, with results reported as P values. Univariate and multivariate analyses were performed for factors affecting change in RLN diameter. Analyses were performed using SAS Software, version 9.2 (SAS Institute, Cary, NC). A two-sided P # .05 was considered to be statistically significant. RESULTS Of the 3,408 thyroidectomy cases (between 2007 and 2013), 1,926 were bilateral procedures and 1,482 were unilateral procedures, resulting in a total of 5,334 at-risk RLNs (Table I). Of these, 2,650 were left RLNs and 2,684 were right RLNs. A total of 328 cases (between 2009 and 2011) had complete RLN diameter and EMG data from
Serpell et al 1159
IONM (before and after lobectomy), resulting in 519 RLNs in the RLN diameter/EMG cohort. In 3 cases, it was documented that one RLN was not identified (2 bilateral cases and 1 unilateral case). Detailed subgroups (right vs left; bilateral vs unilateral procedures; reoperative cases) are shown in Table I. The mean age of the patients was 54 ± 23 (SD) years, and there was a female predominance (approximately 80%) across all cohorts. Only approximately 5% were reoperative procedures. The majority of the cases had a benign diagnosis on histopathology (2,927; 86%), including toxic and nontoxic multinodular goiter (2,209; 65%), Graves’ disease (253; 7.4%) and solitary benign nodules (315; 9%). Malignant diagnoses (481; 14%) included differentiated thyroid carcinoma, medullary thyroid carcinoma, poorly differentiated thyroid carcinoma, and lymphoma (Table II). RLN palsy rates. The RLN palsy rates are summarized in Table III, A and Table III, B. Overall, RLN palsy rate was documented in 1.5% (82) of cases, with temporary palsy contributing 1.3% (68) and permanent palsy contributing 0.3% (14). No difference was seen in the differential palsy rates of the left (1.5%) and right (1.6%) sides (P = .87). In contrast, this was not the case on subgroup analysis. In bilateral procedural cases, the left RLN palsy rate was 0.9% and the right RLN palsy rate was 1.8% (P = .025). For unilateral procedural cases, the left RLN palsy rate was 3%, and the right RLN palsy rate was 1.1% (P = .007). A greater rate of left RLN palsy was recorded in unilateral cases (3%) compared with the left RLN palsy rate of bilateral cases (0.9%; P < .001). In contrast, the right-sided RLN palsy rates were similar whether the procedure was bilateral (1.8%) or unilateral (1.1%). Twelve of 14 permanent RLN palsies occurred on the right side, 10 of which were in cases of bilateral procedures, and 2 in cases of unilateral procedures. There was 1 case of permanent palsy in each of bilateral and unilateral cases on the left side. RLN diameter and conductance. Overall, the mean RLN diameter increased from 1.6 ± 0.1 mm to 2.4 ± 0.1 mm during the procedure (P < .001). The diameter of the RLN was larger on the right side both on initial identification (left 1.6 ± 0.01 mm vs right 1.7 ± 0.1 mm, P = .0012), as well as after completion of lobectomy (left 2.3 ± 0.1 mm vs right 2.5 ± 0.1 mm, P < .001). The diameter of both left and right RLNs increased during the course of the dissection (P < .001 for both sides; Table IV). Although there appeared to be a trend toward a greater increase in
1160 Serpell et al
Surgery November 2014
Table I. Case mix showing the study populations and demographics Cases for RLN palsy rate 2007–2013
Operations RLN at risk L RLN at risk R RLN at risk Mean age (SD) Female % Redo cases %
Cases for RLN diameter and EMG* 2009–2011
All
Bilat
Unilat
All
Bilat
Unilat
3,408 5,334 2,650 2,684 54 (23) 80.5 5.0
1,926 3,852 1,926 1,926 54 (16) 82.1 1.5
1,482 1,482 724 758 54 (20) 78.5 9.7
328 519 260 259 55 (16) 81.9 6.4
191 382 191 191 57 (15) 82.2 0.0
137 137 69 68 53 (15) 81.0 15.3
*The cases for RLN diameter and EMG are a subgroup of the cases for RLN palsy rate. bilat, Bilateral thyroid surgery including reoperative cases; EMG, electromyogram; L, left; R, right; RLN, recurrent laryngeal nerve; redo, reoperative; SD, standard deviation; unilat, unilateral thyroid surgery including reoperative cases.
Table II. Pathology of the cases for each of the study cohorts
Benign MNG GD Solitary nodule Other benign Malignant Total
Cases for RLN palsy
Cases for RLN diameter
2,927 2,209 253 315 150 481 3,408
295 251 21 16 7 33 328
GD, Graves’ disease; MNG, multinodular goiter; RLN, recurrent laryngeal nerve.
absolute size on the right side (left 0.7 mm vs right 0.8 mm, P = .085), the relative increases from the starting diameter were the same bilaterally (left 1.5-fold vs right 1.5-fold, P = .48). In the 8 RLNs that led to a postoperative palsy in this subgroup, the mean diameter at the start of the procedure (palsy 1.6 ± 0.1 mm vs no palsy 1.6 ± 0.1 mm, P = .61) and the increase in diameter (palsy 0.8 ± 0.2 mm vs no palsy 0.7 ± 0.1 mm, P = .79) were both similar to the rest of the cohort. The EMG amplitudes of the left RLNs increased from 634 ± 33 mV to 747 ± 37 mV (P < .001), whereas on the right RLNs, EMG amplitudes increased from 548 ± 32 mV to 575 ± 28 mV (P = .24) between the first and second measurements. The left RLN amplitude increase was 101 ± 26 mV, significantly more than the increase of 26 ± 22 mV on the right side. In the 8 RLNs that led to a postoperative palsy, the EMG amplitude decreased nonsignificantly (315–265 mV, P = .41), whereas for the overall group, EMG amplitude increased with the swelling of the nerves (591–661 mV, P < .001). The mean distance from the inferior border of cricopharyngeus
muscle where RLN diameter was measured was 9.7 ± 0.5 mm. The time between the first and second measurements was 21 ± 1 minutes. The left-right RLN diameter discrepancy was especially pronounced in females (left 1.6 ± 0.1 mm vs right 1.7 ± 0.1 mm, P = .005). There was also a trend toward a larger RLN in males. The mean diameter preoperatively for males was 1.7 ± 0.1 mm, compared with 1.6 ± 0.1 mm for females (P = .07). The diameter increase was also larger in males (0.9 ± 0.1 mm), compared to females (0.7 ± 0.1 mm; P = .007). Increasing age, male sex, smaller initial diameter, and increasing EMG amplitude difference were all significantly associated with increasing RLN swelling during dissection, both on univariate and multivariate models (Table V). Malignant, reoperative, and retrosternal cases did not show a greater increase in nerve diameter increase. DISCUSSION In this study, the left and right RLN palsy rates were determined in a cohort of 3,408 thyroidectomy cases with 5,334 RLNs at risk. Recurrent laryngeal nerve palsy occurred more frequently on the right in bilateral procedures, whereas in unilateral procedures, left RLN palsy occurred more frequently. The RLN diameter was significantly smaller on the left side, and significant increase in diameter occurred during the course of the procedures. The EMG amplitude increased in proportion with the RLN diameter, except in cases of RLN palsy, where the EMG decreased despite increasing diameter. Current published data would suggest that most RLN palsies occur despite visually intact nerves.8,9,19,20 The majority of RLN palsies are temporary and resolve within weeks. Chiang et al reported peak recovery at 6 weeks with 97% recovery
Surgery Volume 156, Number 5
Serpell et al 1161
Table IIIA. Recurrent laryngeal nerve palsy rates, expressed in number and percentage of RLNs
III
*Of the palsies, there were more permanent palsies on the right and more temporary palsies on the left. bilat, Bilateral cases; L, left; N, number; R, right; RLN, recurrent laryngeal nerve; temp, temporary; perm, permanent; unilat, unilateral cases.
by 9 weeks, whereas Sancho et al noted 20 of 33 RLN palsies had recovered by 4 weeks.9,21 Both the visually intactness of the injured nerve and the rapid resolution point toward neurapraxia as the underlying cause of most RLN injuries. Conversely, if the injuries were due to axonotmesis, the recovery would
be much slower. In addition, while uncommon, transection, crushing, and compression injuries causing neurotmesis can be diagnosed usually at the time of injury. Despite these observations, the underlying mechanism of neurapraxia in thyroid surgery has not been elucidated adequately.
1162 Serpell et al
Surgery November 2014
Table IIIB. Palsy rates from Table IIIA, presented pictorially, demonstrating the P values between left-right comparisons, and between unilateral and bilateral procedures on each side
bilat, Bilateral thyroid surgery including reoperative cases; L, left; R, right; RLN, recurrent laryngeal nerve; unilat, unilateral thyroid surgery including reoperative cases.
Table IV. Change in RLN diameter during dissection
Initial diameter, mm (SE) Diameter after dissection, mm (SE) Absolute increase in diameter, mm (SE) Relative increase in diameter, fold (SE) Diameter increase P value Initial EMG amplitude, mV (SE) EMG amplitude after dissection, mV (SE) Absolute increase in amplitude, mV (SE) Relative increase in amplitude, fold (SE) Amplitude increase P value
Both sides N = 519
L RLN N = 260
R RLN N = 259
L vs R P value
1.6 (0.1) 2.4 (0.1) 0.7 (0.1) 1.5 (0.1)
Introduction. Recurrent laryngeal nerve (RLN) palsy is a devastating complication of thyroidectomy. Although neurapraxia is thought to be the most common cause, the underlying mechanisms are poorly understood. The objectives of this study were to examine the differential palsy rates between the left and right RLNs, and the role of intraoperative nerve swelling as a risk factor of postoperative palsy. Methods. Thyroidectomy data were collected, including demographics, change in RLN diameter, and RLN electromyographic (EMG) reading. Left and right RLNs, as well as bilateral and unilateral subgroup analyses were performed. Results. A total of 5,334 RLNs were at risk in 3,408 thyroidectomies in this study. The overall RLN palsy rate was 1.5%, greater on the right side than the left for bilateral cases (P = .025), and greater on the left side than the right for unilateral cases (P = .007). In a subgroup of 519 RLNs, the diameter and EMG amplitude were measured. The RLN diameter increased by approximately 1.5-fold (P < .001), and corresponded to increased EMG amplitude (P = .01) during the procedure. The diameter of the right RLN was larger than the left RLN, both at the beginning and end of the dissection (P = .001). Conclusion. The right-left differential rates of post-thyroidectomy RLN palsy seemed to be due in part to differential RLN diameters, with stretch having a more deleterious effect on RLNs with a smaller diameter; also, edema as a result of stretch might be an underlying mechanism for postoperative neurapraxia and palsy. Thyroid surgeons should be aware of the different vulnerabilities of each RLN and develop practices to avoid iatrogenic injury. (Surgery 2014;156:1157-66.) From the Monash University Endocrine Surgery Unit and Alfred Hospital,a and Department of Epidemiology and Preventative Medicine,b Monash University, Melbourne, Victoria, Australia
THYROIDECTOMY IS A COMMON OPERATION and one of its most devastating complications is recurrent laryngeal nerve (RLN) palsy.1-3 In a recent systematic review, the mean rates of temporary and permanent RLN palsy were 9.8% (range 1.4–38.4%), and 2.3% (range 0–18.6%), respectively.4 Functional consequences of RLN palsy range from dysphonia, weakened cough, and predisposition to aspiration, to life-threatening airway obstruction. A key aim of thyroid surgery is to avoid RLN palsy; to achieve this with a precise surgical technique, the thyroid surgeon needs a detailed Authors are listed in descending order of contribution. The authors thank their database manager Melissa Vereker for the extraction of the data from the database. Accepted for publication July 17, 2014. Reprint requests: Jonathan W. Serpell, MBBS, MD, MEd, FRACS, FACS, The Alfred Hospital, Commercial Road, Prahran, Victoria 3181, Australia. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2014.07.018
understanding of the surgical anatomy and neurophysiology of the RLN and to be aware of the clinical associations and the pathophysiologic causes of this complication.5 Anatomically, the left and right RLNs are different. The left is longer, usually 12 cm from where it winds around the ligamentum arteriosum to the larynx, whereas the right RLN is usually 6 cm from where it winds around the subclavian artery to the larynx. The left RLN tends to run within the tracheoesophageal groove, whereas the right RLN tends to run a more anterior and oblique course to the tracheoesophageal groove.6 Pathophysiologically, RLN palsy can be caused by stretching, heat, compression, devascularization, or transection, which results in neurapraxia, axonotmesis, or neurotmesis.7 Neurapraxia appears to be the most common mode of injury because most RLN injuries are not recognized at the time of operation with the nerves remaining visually intact. Most RLN palsies are temporary and resolve quickly.8,9 Electromyography (EMG) is a useful tool to assess nerve integrity intraoperatively because injuries SURGERY 1157
1158 Serpell et al
other than nerve transection may not be recognizable visually. Although the underlying cause of the neurapraxia is unclear, it has been reported in animal models that excessive tension can cause ischemia and myelin damage.10,11 It has also been reported that RLN diameter increases during thyroidectomy, presumably because of edema, and may be related to the subsequent neurapraxia.12 Therefore, we postulate that tension from intraoperative stretching can lead to decreased blood flow, and the subsequent edema perpetuates the ischemia and causes RLN neurapraxia in neck surgery. Asymmetric RLN anatomy modulating the tension from retraction was postulated as the reason for different rates of palsy between the left and right RLN after anterior cervical laminectomy.13,14 In the large number of thyroidectomy series reporting RLN palsy rates, only 2 studies in the past 5 decades have included the left-right breakdown; they all reported higher palsy rates on the right.15,16 At present, the underlying reasons for the discrepant RLN palsy rates between the left and right sides are poorly understood. There are insufficient data to support the role of the anatomic asymmetry in RLNs to explain the differential rates of right- versus left-sided RLN palsy. Therefore, the aims of this study were to: 1. Compare the left- and right-sided RLN palsy rates after thyroidectomy in a single surgical unit; 2. Investigate the correlation of RLN diameters to the differential palsy rates; and 3. Investigate the interaction between intraoperative RLN swelling and EMG changes, and their role in postoperative RLN palsy.
PATIENTS AND METHODS This retrospective cohort study was approved by the Human Research and Ethics Committee of The Alfred Hospital, Melbourne, Australia. Cases of thyroid procedures between 2007 and 2013 were recruited from the Monash University Endocrine Surgery Unit database. The information from this database was prospectively collected and audited regularly by the database manager. The patients for the measurements of RLN diameter and EMG were recruited between 2009 and 2011. Of the eligible cases, 6 were excluded from the study because of tumor invasion leading to intentional RLN sacrifice (4 cases) or incomplete documentation of the essential parameters (2 cases). Data were collected regarding the case demographics, RLN function pre- and post-thyroidectomy, change
Surgery November 2014
in RLN diameter during dissection, and functional status of the RLN according to the intraoperative neuromonitor (IONM) EMG reading. All procedures were performed under general anaesthesia with 4 mg of intravenous dexamethasone administered routinely on induction. All procedures were performed by 1 of 4 surgeons in the unit adhering to a standard method described previously.17 The RLN was first identified approximately where it was crossed by the inferior thyroid artery. It was then exposed widely by ligation of the tertiary branches of the inferior thyroid artery, resulting in gradual anteromedial rotation of the thyroid lobe.1-3,17 All surgeons used vessel-sealing devices; no thermal devices were used within 1 cm of the RLN. Hemostasis in the vicinity of the RLN was achieved by liga-clips, ligatures, or sutures. Results were analyzed as a unit; therefore, differential results between surgeons or thermal devices were not available. Bilateral procedures were defined as procedures involving both thyroid lobes in the one operative episode, therefore putting both RLNs at risk, and consisted mainly of total thyroidectomy. Bilateral reoperative procedures (eg, bilateral recurrence after previous bilateral subtotal thyroidectomy) were also included as bilateral procedures. Unilateral procedures mainly consisted of hemithyroidectomy (or lobectomy), but also included unilateral reoperative procedures. Sequential lobectomies, such as a diagnostic hemithyroidectomy followed by completion thyroidectomy in a subsequent operative episode were considered as two unilateral procedures, if both were captured during the study period. On initial identification, the diameter of the RLN was measured with vernier calipers with a resolution of 0.1 mm. The measurement was repeated just before the lobe was detached from the trachea at Berry’s ligament. The second measurement was made at the same point of the nerve where the first measurement was taken. The distance from inferior border of cricopharyngeus muscle was used as a reference to the position of the first measurement. If branching of the RLN was present, the diameter of the main trunk was measured. The IONM was performed with the NIM Nerve Monitoring System (Medtronics, MN), using the accompanying handheld neurostimulatory probe and the EMG endotracheal tube. The stimulation of the probe was set at 1.0 mA, which resulted in supramaximal stimulation of the RLN. The vagus nerve was stimulated routinely at the beginning and end of the case. If a muscle relaxant was used
Surgery Volume 156, Number 5
on induction of general anesthesia, either the muscle relaxant was reversed or the return of muscular activity was confirmed before EMG recording. Vocal cord adduction was recorded by the EMG endotracheal tube, and the threshold sensitivity was set at 100 mV.4,18 Abduction of the vocal cord was assessed by the laryngeal twitch technique, with palpation of contraction by the posterior cricoarytenoid muscle.5,18 The maximum amplitude achieved was recorded in the database. All patients had pre- and postoperative flexible fiberoptic laryngoscopy by a surgeon independent of the treating surgeon to assess vocal cord function. The postoperative assessment was performed one day after the operation but earlier if there were concerns regarding airway patency in the recovery room. Vocal cord assessment at this time point was the routine of the unit to maximize capture of temporary palsies and to minimize noncompliance after discharge. Subsequent vocal cord assessments were performed if an abnormality was detected at the initial postoperative assessment. The patient was monitored until resolution of the abnormality or definitive management of the palsy was instituted. A RLN palsy was regarded as permanent if it persisted over 6 months. Statistical analysis. Because some patients were included more than once if they had repeat operations, cases, procedures, or number of RLNs are referred to in this manuscript, instead of number of patients. All data were assessed initially for normality. Incidence rates were compared using the chi-squared test for equal proportions. Changes in RLN diameter were assessed by paired t tests and reported as means and standard errors (SE). All mean values are reported with SE values, unless stated otherwise. To account for repeated measures, the relationships between increase in RLN diameter and other parameters were assessed by generalized linear modelling, with results reported as P values. Univariate and multivariate analyses were performed for factors affecting change in RLN diameter. Analyses were performed using SAS Software, version 9.2 (SAS Institute, Cary, NC). A two-sided P # .05 was considered to be statistically significant. RESULTS Of the 3,408 thyroidectomy cases (between 2007 and 2013), 1,926 were bilateral procedures and 1,482 were unilateral procedures, resulting in a total of 5,334 at-risk RLNs (Table I). Of these, 2,650 were left RLNs and 2,684 were right RLNs. A total of 328 cases (between 2009 and 2011) had complete RLN diameter and EMG data from
Serpell et al 1159
IONM (before and after lobectomy), resulting in 519 RLNs in the RLN diameter/EMG cohort. In 3 cases, it was documented that one RLN was not identified (2 bilateral cases and 1 unilateral case). Detailed subgroups (right vs left; bilateral vs unilateral procedures; reoperative cases) are shown in Table I. The mean age of the patients was 54 ± 23 (SD) years, and there was a female predominance (approximately 80%) across all cohorts. Only approximately 5% were reoperative procedures. The majority of the cases had a benign diagnosis on histopathology (2,927; 86%), including toxic and nontoxic multinodular goiter (2,209; 65%), Graves’ disease (253; 7.4%) and solitary benign nodules (315; 9%). Malignant diagnoses (481; 14%) included differentiated thyroid carcinoma, medullary thyroid carcinoma, poorly differentiated thyroid carcinoma, and lymphoma (Table II). RLN palsy rates. The RLN palsy rates are summarized in Table III, A and Table III, B. Overall, RLN palsy rate was documented in 1.5% (82) of cases, with temporary palsy contributing 1.3% (68) and permanent palsy contributing 0.3% (14). No difference was seen in the differential palsy rates of the left (1.5%) and right (1.6%) sides (P = .87). In contrast, this was not the case on subgroup analysis. In bilateral procedural cases, the left RLN palsy rate was 0.9% and the right RLN palsy rate was 1.8% (P = .025). For unilateral procedural cases, the left RLN palsy rate was 3%, and the right RLN palsy rate was 1.1% (P = .007). A greater rate of left RLN palsy was recorded in unilateral cases (3%) compared with the left RLN palsy rate of bilateral cases (0.9%; P < .001). In contrast, the right-sided RLN palsy rates were similar whether the procedure was bilateral (1.8%) or unilateral (1.1%). Twelve of 14 permanent RLN palsies occurred on the right side, 10 of which were in cases of bilateral procedures, and 2 in cases of unilateral procedures. There was 1 case of permanent palsy in each of bilateral and unilateral cases on the left side. RLN diameter and conductance. Overall, the mean RLN diameter increased from 1.6 ± 0.1 mm to 2.4 ± 0.1 mm during the procedure (P < .001). The diameter of the RLN was larger on the right side both on initial identification (left 1.6 ± 0.01 mm vs right 1.7 ± 0.1 mm, P = .0012), as well as after completion of lobectomy (left 2.3 ± 0.1 mm vs right 2.5 ± 0.1 mm, P < .001). The diameter of both left and right RLNs increased during the course of the dissection (P < .001 for both sides; Table IV). Although there appeared to be a trend toward a greater increase in
1160 Serpell et al
Surgery November 2014
Table I. Case mix showing the study populations and demographics Cases for RLN palsy rate 2007–2013
Operations RLN at risk L RLN at risk R RLN at risk Mean age (SD) Female % Redo cases %
Cases for RLN diameter and EMG* 2009–2011
All
Bilat
Unilat
All
Bilat
Unilat
3,408 5,334 2,650 2,684 54 (23) 80.5 5.0
1,926 3,852 1,926 1,926 54 (16) 82.1 1.5
1,482 1,482 724 758 54 (20) 78.5 9.7
328 519 260 259 55 (16) 81.9 6.4
191 382 191 191 57 (15) 82.2 0.0
137 137 69 68 53 (15) 81.0 15.3
*The cases for RLN diameter and EMG are a subgroup of the cases for RLN palsy rate. bilat, Bilateral thyroid surgery including reoperative cases; EMG, electromyogram; L, left; R, right; RLN, recurrent laryngeal nerve; redo, reoperative; SD, standard deviation; unilat, unilateral thyroid surgery including reoperative cases.
Table II. Pathology of the cases for each of the study cohorts
Benign MNG GD Solitary nodule Other benign Malignant Total
Cases for RLN palsy
Cases for RLN diameter
2,927 2,209 253 315 150 481 3,408
295 251 21 16 7 33 328
GD, Graves’ disease; MNG, multinodular goiter; RLN, recurrent laryngeal nerve.
absolute size on the right side (left 0.7 mm vs right 0.8 mm, P = .085), the relative increases from the starting diameter were the same bilaterally (left 1.5-fold vs right 1.5-fold, P = .48). In the 8 RLNs that led to a postoperative palsy in this subgroup, the mean diameter at the start of the procedure (palsy 1.6 ± 0.1 mm vs no palsy 1.6 ± 0.1 mm, P = .61) and the increase in diameter (palsy 0.8 ± 0.2 mm vs no palsy 0.7 ± 0.1 mm, P = .79) were both similar to the rest of the cohort. The EMG amplitudes of the left RLNs increased from 634 ± 33 mV to 747 ± 37 mV (P < .001), whereas on the right RLNs, EMG amplitudes increased from 548 ± 32 mV to 575 ± 28 mV (P = .24) between the first and second measurements. The left RLN amplitude increase was 101 ± 26 mV, significantly more than the increase of 26 ± 22 mV on the right side. In the 8 RLNs that led to a postoperative palsy, the EMG amplitude decreased nonsignificantly (315–265 mV, P = .41), whereas for the overall group, EMG amplitude increased with the swelling of the nerves (591–661 mV, P < .001). The mean distance from the inferior border of cricopharyngeus
muscle where RLN diameter was measured was 9.7 ± 0.5 mm. The time between the first and second measurements was 21 ± 1 minutes. The left-right RLN diameter discrepancy was especially pronounced in females (left 1.6 ± 0.1 mm vs right 1.7 ± 0.1 mm, P = .005). There was also a trend toward a larger RLN in males. The mean diameter preoperatively for males was 1.7 ± 0.1 mm, compared with 1.6 ± 0.1 mm for females (P = .07). The diameter increase was also larger in males (0.9 ± 0.1 mm), compared to females (0.7 ± 0.1 mm; P = .007). Increasing age, male sex, smaller initial diameter, and increasing EMG amplitude difference were all significantly associated with increasing RLN swelling during dissection, both on univariate and multivariate models (Table V). Malignant, reoperative, and retrosternal cases did not show a greater increase in nerve diameter increase. DISCUSSION In this study, the left and right RLN palsy rates were determined in a cohort of 3,408 thyroidectomy cases with 5,334 RLNs at risk. Recurrent laryngeal nerve palsy occurred more frequently on the right in bilateral procedures, whereas in unilateral procedures, left RLN palsy occurred more frequently. The RLN diameter was significantly smaller on the left side, and significant increase in diameter occurred during the course of the procedures. The EMG amplitude increased in proportion with the RLN diameter, except in cases of RLN palsy, where the EMG decreased despite increasing diameter. Current published data would suggest that most RLN palsies occur despite visually intact nerves.8,9,19,20 The majority of RLN palsies are temporary and resolve within weeks. Chiang et al reported peak recovery at 6 weeks with 97% recovery
Surgery Volume 156, Number 5
Serpell et al 1161
Table IIIA. Recurrent laryngeal nerve palsy rates, expressed in number and percentage of RLNs
III
*Of the palsies, there were more permanent palsies on the right and more temporary palsies on the left. bilat, Bilateral cases; L, left; N, number; R, right; RLN, recurrent laryngeal nerve; temp, temporary; perm, permanent; unilat, unilateral cases.
by 9 weeks, whereas Sancho et al noted 20 of 33 RLN palsies had recovered by 4 weeks.9,21 Both the visually intactness of the injured nerve and the rapid resolution point toward neurapraxia as the underlying cause of most RLN injuries. Conversely, if the injuries were due to axonotmesis, the recovery would
be much slower. In addition, while uncommon, transection, crushing, and compression injuries causing neurotmesis can be diagnosed usually at the time of injury. Despite these observations, the underlying mechanism of neurapraxia in thyroid surgery has not been elucidated adequately.
1162 Serpell et al
Surgery November 2014
Table IIIB. Palsy rates from Table IIIA, presented pictorially, demonstrating the P values between left-right comparisons, and between unilateral and bilateral procedures on each side
bilat, Bilateral thyroid surgery including reoperative cases; L, left; R, right; RLN, recurrent laryngeal nerve; unilat, unilateral thyroid surgery including reoperative cases.
Table IV. Change in RLN diameter during dissection
Initial diameter, mm (SE) Diameter after dissection, mm (SE) Absolute increase in diameter, mm (SE) Relative increase in diameter, fold (SE) Diameter increase P value Initial EMG amplitude, mV (SE) EMG amplitude after dissection, mV (SE) Absolute increase in amplitude, mV (SE) Relative increase in amplitude, fold (SE) Amplitude increase P value
Both sides N = 519
L RLN N = 260
R RLN N = 259
L vs R P value
1.6 (0.1) 2.4 (0.1) 0.7 (0.1) 1.5 (0.1)
