Investigation of the Impact of Thyroid Surgery on Vocal Tract Steadiness

Results Mixed analysis of variance (ANOVA) for all acoustic parameters revealed that the χ F 2 FF showed a significant “time” main effect ( F (1,22) = 7.196, P = 0.014, partial η 2 = 0.246) and a significant “time by group interaction” effect ( F (1,22) = 8.036, P = 0.010, η p 2 = 0.268), with changes over time for the thyroid group but not for the controls. Similarly, mean χ F 2 FFR showed a similar significant “time” main effect ( F (1,22) = 6.488, P = 0.018, η p 2 = 0.228) and a “time by group interaction” effect ( F (1,22) = 7.134, P = 0.014, η p 2 = 0.245). Conclusions This work suggests that thyroid surgery produces a significant reduction in vocal tract stability in contrast to the controls. This noninvasive measurement offers a potential instrument to investigate the functional implications of any disturbance that thyroid surgery may have on pharyngeal innervations. Key Words Thyroid surgery Postoperative dysphonia Vocal tract steadiness Formant frequency fluctuation Introduction Approximately 14% of patients experience voice symptoms that persist beyond 6 months after thyroid surgery. Historically, these symptoms have been accepted as resulting from an iatrogenic injury to the laryngeal nerves during surgery. However, since the 1950s and 1960s, several surgeons have reported postsurgical voice changes in patients who have no demonstrable evidence of nerve injury. 1,2 The predominant feature of these patients is a loss of vocal range and vocal fatigue. More recently, numerous studies have used a variety of acoustical and electromyography (EMG) techniques to analyze these patients without producing a consistent finding to explain these symptoms. 3–6 A few studies also investigated these patients for swallowing complaints. These articles report that those with the most pronounced voice symptoms tend to have swallowing complaints also. 7,8 This led to the speculation that there may be a common anatomical cause for these symptoms. Several anatomic studies have shown the presence of extralaryngeal branches of the recurrent laryngeal nerve that often anastomose with other cervical nerves. 9,10 Indeed, there is evidence that small branches of the recurrent nerve participate with other cervical nerves to form a perivisceral nerve plexus that plays a part in the regulation of the autonomic, motor, and sensory function of both the larynx and pharynx. 11 Therefore, despite apparent preservation of vocal cord function, injury to this plexus may cause minor postoperative symptoms. Indeed, this anatomic variation of cervical nerves may be responsible for the wide array of postoperative voice alterations observed. To investigate this, we used a technique that has been used over the past 30 years primarily to assess patients with neurological disorders of the larynx, vocal tract steadiness. 12 This noninvasive acoustic measurement is based on the Source and Filter theory, which states that changes in vocal tract configuration can be measured as changes in the formant frequencies. In particular, the sequential segment-to-segment fluctuations of the second formant of a sustained vowel can be used as an index of the degree of the moment-to-moment fluctuations in vocal tract configuration. Materials and Methods Ethical approval for the study was sought and granted by the Ethics and Research Committee of the Royal Victoria Eye & Ear Hospital in January 2007. All patients in the study were over 18 years of age with no previous history of any voice disorder or potentially related condition of the head and neck. Once a thyroidectomy patient had been enrolled in the study, a patient of similar gender and age undergoing an elective ear or nasal operation was also enrolled in the study as part of the matched controls. All thyroid surgeries were performed by one surgeon (Conrad I. Timon) with the use of the Neurosign 100 intraoperative nerve monitor (Magstim Co., Ltd. Wales) with matching endotracheal tube. In all cases, the superior and recurrent laryngeal nerves were stimulated once to confirm their electrophysiological integrity at the end of the operation. The first stimulation is always set at 0.2 mA for the superior laryngeal nerve and 0.5 mA for the recurrent laryngeal nerve to avoid the risk of inducing fatigue to the nerve by excessive electrical stimulation. Voice recordings Before all vocal recording, patients had a flexible laryngoscopy examination to out rule any laryngeal pathology or paresis. This consisted of a flexible nasopharyngoscope examination without any assessment of sensation. During this examination, specific anatomic findings were recorded and specific vocal tasks were assessed. Specifically, patients were viewed during quiet respiration, during rapid repeated adductions, during a vigorous inspiration through the nose, and finally, the patients were asked to perform a glide to assess the cricothyroid function. If the patients had any evidence of impaired function of these tasks preoperatively, they were excluded from the study; whereas, any of these findings in the postoperative assessment were recorded. Patients who had any of these findings were excluded from the study. All recordings were made using the Laryngograph (Laryngograph Ltd., London, UK). Each patient was seated and a head-worn high-fidelity microphone (Labtec XC47, www.labtec.com ) was positioned at a constant 20 cm from the lips. Patients were asked to take a deep breath and to sustain the vowel /a/ at a conversational level of loudness for as long, steadily, and clearly as possible. In addition, each subject was instructed to maintain the same vowel quality throughout the entire production. After practice, they were asked to do this three times. A 3-second segment from the middle portion of the longest production was used for analysis. The recordings were taken preoperatively and again on the second postoperative day. When these recordings were done postoperatively, patients were questioned about the amount of pain they were experiencing using a visual analog scale from 1 to 10. If patients were experiencing pain of value greater than 1, they were given analgesia as an attempt to remove the effect of pain on the patients' voicing and articulation. Acoustical analysis Vocal tract stability was examined by measuring the formant frequency fluctuation (FFF). The second formant frequencies ( F 2 ) were calculated using the autocorrelation method of linear prediction based on the work of Gerratt. 12 Ten consecutive nonoverlapping 100-millisecond segments were used from the selected segment. Peak values were considered invalid if their bandwidth greater than 300 Hz. The segmental F 2 values were used to compute four measures of F 2 : (1) mean, (2) standard deviation (SD), (3) mean FFFs ( F 2 FF), and (4) F 2 FF ratio ( F 2 FFR) in the same manner as described in the original work of Gerratt. 12 The purpose of calculating the FFFR is to attempt to evaluate the different vocal tract characteristics of males and females, and is in accordance with previous publications. 12 Statistical analysis All the data was recorded in an Excel (Microsoft corporation, USA) spreadsheet and exported into the SPSS Statistical Programm (Chicago, IL). The experimental design yielded two distinct groups. Differences between the groups were examined using analysis of variance (ANOVA) for age, Mann-Whitney U test for duration of operation and the Fisher's exact test for gender. All parameters of vocal tract steadiness were analyzed with factorial repeated-measures ANOVA (2×2 design). For significance, P values were set at less than 0.005. Results A total of 12 patients were successfully recruited into the thyroid group. There were two males and 10 females. The mean age was 48.75 years and SD = 14.30 years (range = 28–70 years). Five patients had a hemithyroidectomy, two of whom had partial division of the sternothyroid muscle and three had retraction only. Five patients had a total thyroidectomy; three of these patients had partial muscle division on both sides, one patient had division on the left, and the final patient had retraction of the muscle only. Two patients had an endoscopic parathyroidectomy and neither required any division of the strap muscles. All patients had both superior and recurrent nerves identified and confirmed by the intraoperative nerve probe at the basal stimulation threshold level. In the control group, there were three males and nine females. The mean age was 47.08 years and SD = 16.06 years (range = 26–69 years). Five patients had sinus surgeries and seven patients had ear surgeries. There was no statistically significant difference between the two groups on age ( F (1,22) = 0.072, P = 0.791), gender (Fisher's test, P = 0.55; Cramer's V = 0.103) and duration of operation (Mann-Whitney U = 60, z = 0.799, P = 0.424). ANOVA for the mean F 2 FF showed a significant “time” main effect ( F (1,22) = 7.196, P = 0.014, η p 2 = 0.246), with the postoperative values being higher (mean = 24.332 Hz, standard error = 2.089) than the preoperative ones (mean = 33.911, standard error = 3.660). Analysis also revealed a significant “time by group interaction” effect ( F (1,22) = 8.036, P = 0.010, η p 2 = 0.268). Within the thyroid group, the postoperative mean F 2 FF scores were significantly higher (mean = 43.00 Hz, standard error = 5.176) than the preoperative values (mean = 23.30 Hz, standard error = 2.954). However, there was no significant difference between the preoperative values (mean = 25.36 Hz, standard error = 2.954) and the postoperative values (mean = 24.82 Hz, standard error = 5.176) of mean F 2 FF in the control group. Furthermore, the preoperative values for the control group and the thyroid group did not differ ( F (1,22) = 0.243, P = 0.627, η p 2 = 0.011), but the postoperative values were significantly different ( F (1,22) = 6.171, P = 0.021, η p 2 = 0.219), ( Figure 1 ). The F 2 FFR displayed a parallel pattern as the mean F 2 FF. There was a similar “time” main effect ( F (1,22) = 6.488, P = 0.018, η p 2 = 0.228) and “time by group interaction” effect ( F (1,22) = 7.134, P = 0.014, η p 2 = 0.245). Within the thyroid group, the postoperative F 2 FFR scores were significantly higher (mean = 16.85, standard error = 2.087) than preoperative F 2 FFR scores (mean = 9.13, standard error = 1.150). However, there was no significant difference between the preoperative (mean = 10.48, standard error = 1.150) and the postoperative values (mean = 10.29, standard error = 2.087) in the control group. Furthermore, the preoperative values for F 2 FFR of the control group and thyroid group did not differ ( F (1,22) = 0.689, P = 0.416, η p 2 = 0.030), but at postoperative measurement, they are significantly different ( F (1,22) = 4.936, P = 0.037, η p 2 = 0.183) as seen in Figure 2 . Discussion The original application of this noninvasive method was carried out by Gerratt in 1983. 12 He sampled 78 25.6-millisecond segments that overlapped every 12.8 seconds to capture smaller variations in F 2 . However, of the other studies that have used this technique, no other authors have used such short and overlapping segments. Indeed, Robb et al and Zwirner et al have all chosen to take consecutive 100-millisecond segments in the midportion of the vowel sample. 13,14 As these studies do not appear to have been compromised by this altered sampling rate, we decided to adopt this less-involved method of calculation, and it does not appear to have adversely affected our findings. We observed that F 2 FF and F 2 FFR were significantly altered postoperatively when compared with the preoperative values. No such significant changes were noted within the control group. Preoperatively, there was no difference between the two groups for F 2 FF and F 2 FFR, whereas there was a significant difference in these two parameters between the two groups postoperatively. Therefore, there is a significant increase in F 2 fluctuations caused by thyroid and parathyroid surgeries, which is not witnessed in matched controls. However, one must interpret these findings with some caution for a number of reasons. Firstly, it must be acknowledged that although the group statistics reach significance for F 2 FF and F 2 FFR, the numbers in this study are less than optimal. A further compounding factor of the study's small numbers was that the thyroid group was not homogeneous for any particular thyroid operation. Within the group, five patients had hemithyroidectomies, five had total thyroidectomies, and two had endoscopic parathyroidectomies. On direct inspection of the individual groups, in three of the 12 patients, the F 2 parameters did not appear to have changed by a significant degree. All of these patients underwent different operations, which makes meaningful comment difficult. One had an endoscopic parathyroidectomy, one had an endoscopic hemithyroidectomy, and the third had a conventional total thyroidectomy. To date, the two most postulated theories forwarded toward an explanation for the 14% of patients who experience voice derangements in the absence of a nerve injury are to do with either the concept that removal of the thyroid gland results in fixation of the strap muscles and larynx into a form of complex with reduced laryngeal elevation. The other theory rests around the existence of a plexus of nerves that is responsible for innervations of the laryngopharynx. However, in addition to these, there are a variety of other factors, such as trauma to the larynx or arytenoids during intubations, and possible functional alterations in the vocal tract that arise after surgery for a variety of reasons, including altered lymphatic drainage and emotional factors which have been implicated in this phenomenon. 3–8,15 In relation to the theory of a perivisceral nerve plexus, it is postulated that either or both of the superior and recurrent laryngeal nerves contribute to this via very small branches that may be damaged during the mobilization of the gland at the time of the operation. If the former theory was to be accepted, then one would postulate that a parathyroidectomy should not impact on the voice in this manner as the thyroid gland remains, and should provide the necessary structural support and separation of larynx, trachea, and strap muscles not to account for this phenomenon. On the contrary, if the neural plexus theory is considered, then it is possible that a parathyroidectomy could just as easily interrupt this plexus as a thyroid operation. In all the reported series, only a percentage of patients experience this clinical entity. The authors report that between 30% and 68% have the symptoms in the initial period, whereas that number is down to only 14% of patients at 3 months postsurgery. 3–8 There is considerable variation in the configuration of both the recurrent and superior laryngeal nerves, whereas, to date, there is surprisingly little evidence to suggest that how the strap muscles are managed perioperatively has any impact on voice and/or swallow symptoms postoperatively. 4,16 Specifically, Henry et al analyzed 84 patients (45 had the sternothyroid muscle divided and 39 did not have it divided) and found that the management of the sternothyroid muscle had no impact on the incidence or extent of postoperative voice alterations. 16 Therefore, it may seem more appropriate to accept the concept of a perivisceral nerve plexus injury theory. It is possible that it is only patients with a certain branching pattern and cross-innervation pattern who are susceptible to developing these symptoms. If this were correct, then it would explain the variation seen even in this small study. However, it may also be that these findings are as a result of a complex interaction of factors that a single measurement may not detect. Therefore, to better pursue these two theories, a direct comparison between thyroid and parathyroid surgery patients is required in the future. Finally, it may be that the effects on F 2 parameters are of a transient nature and have little bearing on the patient's clinical picture. Therefore, these measurements need to be taken repeatedly and for at least 3 months after the surgery, in conjunction with both subjective and objective measures of voice and swallow, to better establish if they are clinically pertinent. Conclusion The experiment hypothesis was that thyroid and parathyroid surgeries would have a greater effect on F 2 frequency fluctuation and its F 2 frequency fluctuation ratios than nonthyroid surgery of a similar duration. Mixed ANOVA analysis revealed that there was a significant effect on both variables as a consequence of thyroid and parathyroid surgeries. These findings lend some potentially exciting credentials to the concept that the recurrent and superior laryngeal nerves have clinically significant innervations to the supraglottic motor function in some patients, who could be injured at the time of surgery, thus introducing vocal tract stability as a noninvasive measurement instrument with potential clinical import in surgical patients. Further studies will be required to overcome this work's shortcomings. In particular, larger numbers of patients in the subsets of thyroid and parathyroid surgeries should be followed for at least 3 months, and their vocal tract stability should be compared, with both objective and subjective assessments of both voice and swallow; this is required for a direct and more powerful comparison of the two groups. Acknowledgments This work was conducted as part of a project for MSc in Voice Pathology at University College London. We wish to acknowledge the financial support of the Royal Victoria Eye & Ear Hospital Education and Research Fund for the purchase of all equipments used in this work. References 1 N.S. Painter Results of surgery in the treatment of toxic goitre—review of 172 cases Br J Surg 48 1960 291 296 2 D.F.E. Nash Thyroidectomy and the recurrent laryngeal nerves Lancet ii 1956 632 3 K.H. Hong Y.K. Kim Phonatory characteristics of patients undergoing thyroidectomy without laryngeal nerve injury Otolaryngol Head Neck Surg 117 1997 399 404 4 N.P. McIvor D.J. Flint J. Gillibrand R.P. Morton Thyroid surgery and voice-related outcomes Aust N Z J Surgery 70 2000 179 183 5 D.L. Singara M.R. Montesinos V.A. Tacchi Voice changes after thyroidectomy without recurrent laryngeal nerve injury J Am Coll Surg 199 2004 556 560 6 L. Soylu S. Ozbas H.Y. Uslu S. Kocak The evaluation of the causes of the subjective voice disturbances after thyroid surgery Am J Surg 194 2007 317 322 7 J.A. Pereira M. Girvent J.J. Sancho C. Parada A. Sitges-Serra Prevalence of long-term upper aerodigestive symptoms after uncomplicated bilateral thyroidectomy Surgery 133 2003 318 322 8 C.P. Lombardi M. Raffaelli L. D'Altari M.R. Marchese M. Rigante G. Pauludetti R.B. Bellatone Voice and swallowing changes after thyroidectomy in patients without inferior laryngeal nerve injuries Surgery 140 2006 1026 1034 9 J.L. Steinberg G.J. Khane C.M. Fernandes J.P. Nel Anatomy of the recurrent laryngeal nerve: a redescription J Laryngol Otol 100 1986 919 927 10 V. Schweizer J. Dorfl The anatomy of the inferior laryngeal nerve Clin Otolaryngol 22 1997 362 369 11 H.A. Brok M.P. Copper R.J. Stroeve B.W. Ongerboer de Visser A.J. Verker-van Haagen Evidence for recurrent laryngeal nerve contribution in motor of the human cricopharyngeus muscle Laryngoscope 109 1999 705 708 12 B.R. Gerratt Formant frequency fluctuations as an index of motor steadiness in the vocal tract J Speech Hear Res 26 1983 297 304 13 P. Zwirner T. Murray G.E. Woodson Effects of botulinum toxin on vocal tract steadiness in patients with spasmodic dysphonia Eur Arch Otorhinolaryngol 254 1997 391 395 14 M. Robb M. Blomgren Y. Chen Formant frequency fluctuations in stuttering and nonstuttering adults J Fluency disord 23 1998 73 84 15 A. Stojadinovic A. Shaha R.F. Orlikoff Prospective functional voice assessment in patients undergoing thyroid surgery Ann Surg 236 2002 823 832 16 L.R. Henry N.P. Solomon R. Howard The functional impact on voice of sternothyroid muscle division during thyroidectomy Ann Surg Oncol 15 2008 2027 2033