- Email: [email protected]
Voice quality after narrow-margin laser cordectomy compared with laryngeal irradiation
Voice quality after narrow-margin laser cordectomy compared with laryngeal irradiation KATHELIJNE G. DELSUPEHE, MD, INGE ZINK, PhD, MARYLINE LEJAEGERE, MA, and ROBERT W. BASTIAN, MD,
Chicago, Illinois, and Leuven, Belgium
Voice quality remains the issue often used to support preference for radiotherapy in treatment of early glottic cancer. We therefore conducted a perceptual voice study in 2 groups, one treated with radiotherapy for malignant disease and the other with narrow-margin laser cordectomy for either malignant or extensive benign lesions. Sequential patients, 12 treated with radiotherapy and 30 with CO2 laser excision, were included. Voice samples were recorded before and at intervals after surgery. Ratings of validated judges were used for statistical analysis of various voice characteristics at each time point. Voice deteriorated temporarily after surgery as compared with the radiated group; however, at 6 and 24 months no significant differences were found between the groups. Preferential use of narrow-margin laser cordectomy for appropriate early glottic tumors can be supported not only for oncologic reasons but also on the basis of voice results, cost, and efficiency considerations. (Otolaryngol Head Neck Surg 1999;121:528-33.)
T
he current standard of care for early glottic tumors remains controversial. The 2 competing treatment options, radiotherapy and laser excision, each have reported cure rates of approximately 90%.1-4 Despite similar cure rates, in most places radiotherapy seems to be the preferred option, primarily because of concern that voice quality is poor after cordectomy. Radiotherapy is thought to give better voice results, although no long-term comparative results to detect late deterio-
ration of the voice after radiotherapy have been published.2 In recent years information concerning the microarchitecture and layer structure of the vocal folds has become widely available.5-7 This information has led the senior author (R.W.B.) to use narrow margins, not only at the mucosal periphery of early vocal fold carcinomas but also at the deep plane, because the vocal ligament has been found to be an excellent barrier to spread for many of these tumors. From the beginning of a 10-year experience that began in 1987, it was noted that not only cure rates but also voice results were surprisingly good and satisfying to patients who had undergone this highly conservative kind of surgery. Because of this experience and because of the costeffectiveness and convenience of narrow-margin laser cordectomy as compared with radiotherapy, our first choice for treatment of early glottic carcinoma has become narrow-margin laser cordectomy. However, previous reports notwithstanding concerning the good8-10 or serviceable11 voice results that can be achieved after partial cordectomy, some clinicians continue to believe that unacceptable deterioration of voice quality results from narrow-margin laser cordectomy as compared with radiotherapy.12-14 This continuing belief is perhaps the primary reason that laser excision has not gained more advocates. Therefore this study was conceived to determine whether the surprising equivalence of voice results apparent to us informally would be validated when we performed a formal study to compare the voice results of groups treated with these 2 different modalities. METHODS AND MATERIAL Patient Population
From the Department of Otolaryngology, Loyola University (Drs Delsupehe and Bastian); and the Department of Otolaryngology, University Hospitals Leuven (Drs Delsupehe, Zink, and Lejaegere). Presented at the Annual Meeting of the American Academy of Otolaryngology–Head and Neck Surgery, San Francisco, CA, September 7-10, 1997. Reprint requests: Robert W. Bastian, MD, Department of Otolaryngology–Head and Neck Surgery, Loyola University Medical Center, 2160 S First Ave, Maywood, IL 60153. Copyright © 1999 by the American Academy of Otolaryngology– Head and Neck Surgery Foundation, Inc. 0194-5998/99/$8.00 + 0 23/1/101571 528
Forty-two consecutive patients treated at by a single physician (R.W.B.) at Loyola University Medical Center between January 1991 and March 1994 for known or presumed early stage cancers were studied. The patient and lesion characteristics are listed in Table 1. Treatment Modalities Surgery. Outpatient microlaryngoscopy with the microspot CO2 laser was used in every case. The surgical goal in each case was complete excision of the tumor with a narrow mar-
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
DELSUPEHE et al
Table 1. Patient and lesion characteristics
Age (y) Mean Range Sex (M/F) Smokers (current/ former/never) Drinkers (current/ former/never) Staging Benign Tis T1a T1c T2
Radiotherapy group
Surgery group
63.7 50-83 11/1 2/7/1
65.7 43-80 25/5 6/14/1
3/2/2
8/2/6
0 0 5 6 1
13 5 10 1 1
529
A Significance
NS NS NS
B
NS, Not significant.
gin of normal tissue. In contrast to some recommendations,9,11,15,16 mucosal margins were as little as the width of the laser beam and never more than 2 mm. The deep margin followed the contour of the vocal fold similar to Hirano’s5 description rather than cutting straight across it full thickness, as has been depicted elsewhere16 (Figs 1-4). Radiation. Patients were selected for radiotherapy because of medical considerations, anatomic limitations that prevented a clear view of the lesion during microlaryngoscopy, patient preference (as in a singer who was told that the traditional wisdom was that voice was better after radiotherapy than after surgery), and bilaterality. Radiotherapy was administered with an 8-MV linear accelerator to bilateral ports with field sizes ranging from 5 × 5 to 6 × 6 cm. Wedges were used to establish dose homogeneity. The average duration of radiotherapy was 6 weeks, with a total tumor dose ranging between 60 and 64 Gy. Voice Samples Voices were recorded before therapy, early after treatment, and again at 6 and 24 months. All recordings consisted of 2 sentences of the same phonetically balanced passage.17 Of a potential total of 168 samples (42 patients × 4 time points), 139 available recordings were randomized by a statistician. To test intraindividual variability, 30 duplicate recordings chosen randomly from among the 139 were scattered randomly within the samples, to create a total of 169 samples to be rated by the judges. Voice Characteristics Assessed All recordings were evaluated by 8 judges (4 trained and 4 untrained). The trained judges were all speech pathologists. The following auditory-perceptual voice characteristics were rated: overall voice quality, loudness, air wastage or breathi-
Fig 1. T1a carcinoma of the right vocal fold. A, Small lesion outlined with narrow margin. B, After excision, the vocal ligament was approached but not fully exposed.
ness, fluency, sense of effort, and appropriateness of fundamental frequency. The untrained judges evaluated only the first 3 of these characteristics, whereas the trained judges rated them all. Scoring of overall voice quality, loudness, and fluency were based on the following rating scale: 1, poor; 2, fair; 3, good; and 4, excellent. For air wastage and effort, the following scores were used: 1, severe; 2, moderate; and 3, none. Finally, fundamental frequency (F0) was rated as follows: 0, too high or too low; and 1, acceptable. Statistics To determine whether both treatment groups were similar, we compared patient factors and lesion characteristics for both groups using the Fisher exact test. Interjudge variability was assessed for each voice characteristic at each of the four time points, using the Spearman correlation coefficient. Intrajudge variability was analyzed using κ statistics. For all measurements the weighted κ value was used. The data set for all 8 judges was composed of 6075 judgments. On the basis of interjudge and intrajudge variability results for all 8 judges, perceptual ratings of the 2 most consistent judges (a set of 2028 ratings) were used for further statistical analysis. Ratings of these 2 judges for each of the voice char-
530
Otolaryngology– Head and Neck Surgery November 1999
DELSUPEHE et al
A
B
C
D
Fig 2. T1a carcinoma of the right vocal fold. A, Arrows show approximate margin that will be used for this well-circumscribed lesion. B, A narrow mucosal margin has been made with the laser. C, The deep plane of excision is the vocal ligament, which is clearly exposed. The deep surface of the flap containing the lesion is clean. D, Resulting narrow-margin laser cordectomy defect. Note that excision did not transect the fold but followed the plane of the vocal ligament.
acteristics were compared between treatment modalities by use of the Mann-Whitney U test to seek statistical differences. The Wilcoxon test was used to compare the scores within 1 treatment group at different time points. In addition, both tests were used for subgroup analysis (ie, voice result compared with depth of surgical excision and tumor stage). The graphs were plotted with the mean scores of both judges for each item. In those few instances when the weighted κ value for an individual judgment (of a particular voice characteristic) showed only poor or fair strength of agreement (weighted κ < 0.5), this particular rating was omitted when the graph was plotted.18 RESULTS
Forty-two patients entered the study. Twelve patients underwent radiotherapy, and 30 underwent narrow-margin laser cordectomy. Both groups were comparable with regards to age, sex, alcohol and smoking habits, and preoperative staging (Table 1). In the surgery group 13 patients with benign lesions and 5 with carcinoma in situ were included. These patients were kept in the study because their lesions were extensive and the exci-
sions were true narrow-margin laser cordectomies of magnitudes equal to and sometimes greater than those of many of the carcinomas. All others were invasive carcinomas or carcinoma in situ. Interjudge variability for all 8 judges was mostly very good (r > 0.5).19 Overall, trained judges 2 and 3 correlated best with the other judges (highest r values). Intrajudge variability (weighted κ value) assessment again showed trained judges 2 and 3 to have the best correlations (mean weighted κ values of 0.7 and 0.92, respectively). Before treatment there was no significant difference between the 2 treatment modalities in the ratings of any of the voice characteristics. In other words, before treatment the 2 groups were indistinguishable with respect to not only overall quality but also air wastage, loudness, fluency, effort, and F0. There was an improvement over time in the voice quality of both groups. This improvement reached statistical significance in the surgery group for overall quality (P = 0.03) and F0 within 6 months, and for air wastage (P = 0.029), fluency (P = 0.04), loudness (P = 0.012) by 24 months.
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
A
B
DELSUPEHE et al
531
A
B
Fig 3. Larger T1a cancer of the left vocal fold. A, Note the bulk and convexity of the free margin. B, After excision, which in this case took mucosa, ligament, and a small amount of muscle, leaving muscle exposed (arrow).
In the radiotherapy group there was a statistically significant improvement for air wastage only at 6 months (P = 0.017). In the surgery group there was a temporary worsening of all voice characteristics early after surgery. This turned into a net improvement by 6 months after surgery that was sustained at 24 months. When voice results for both treatment modalities were compared over time (Fig 5), the curves were very similar. Evaluation of all of the voice parameters showed no statistical differences between the radiotherapy group and the surgery group 6 and 24 months after treatment with the Mann-Whitney U test. Subgroup analysis was done for the depth of excision in the surgery group (Fig 6). In group A, a plane within the lamina propria or the curved surface of the ligament was the deep margin (only the mucosa and part or all of the superficial layer of the lamina propria were removed). In group B the muscle was the deep margin (mucosa and ligament removed). In group C the superficial cordectomy included part of the thyroarytenoid muscle.8,20 For all of the voice characteristics there was a clear trend in favor of both groups A and B
Fig 4. T1a lesion of the right vocal fold. A, Note that the tumor crowds but does not cross the anterior commissure (white arrow). It lies against the anterior left vocal fold (black arrows). B, After excision, the internal perichondrium of the thyroid cartilage is visible (large arrow). Although full-thickness muscle is gone in a small area anteriorly, most of the muscle of the fold as a whole remains intact (small arrows).
compared with the deeper excision (group C). Because of the small sample size, this reached significance only for overall voice quality at 6 months (group A vs group C; P = 0.024) (Fig 5). In a second subgroup analysis we compared the voices according to the different stages of the lesions in both treatment groups (Fig 7). The preoperative staging correlated very well with the pretreatment voice quality, which tends to validate the judges’ ratings (larger tumors tended to have worse voices). After treatment in the radiotherapy group, the voices improved irrespective of staging differences. In the surgery group there is a trend toward worse voice quality after removal of bigger tumors. Figure 7 shows an example of the effect of stage on overall voice quality.
532
DELSUPEHE et al
Fig 5. Evolution of overall voice quality at 4 time periods: before treatment, early after treatment, and at 6 and 24 months after surgery or radiotherapy.
Fig 6. Evolution of overall voice quality, surgery group only, according to depth of excision.
DISCUSSION
This study evaluates various voice characteristics before and after treatment of early glottic tumors with either radiation therapy or narrow-margin laser cordectomy. Despite the current interest in so-called “objective” measures of vocal output (eg, aerodynamic and acoustic analysis) the use of only auditory perception21 of validated judges for this study was intentional. This is because we were specifically interested in the “real life” question: Can trained and untrained judges distinguish between treatment groups by listening carefully to the voice result? The basic answer is that even the best 2 of 8 judges were unable to do so once surgical healing was complete. Several findings tend to validate the methods and results achieved here. First, with respect to the judges, there was a high degree of interjudge and intrajudge consistency. Second, the voice characteristics of the 2 groups were judged to be the same before treatment to a statistically significant degree, suggesting that they were comparable groups in terms of pretreatment phonatory disturbance. Third, there was a clear and similar trend of voice improvement for most of the characteristics over time. Finally, results in the subgroup
Otolaryngology– Head and Neck Surgery November 1999
Fig 7. Evolution of voice quality in the surgery group according to substage.
analysis were exactly as would be expected (eg, voice results for larger excisions tended to be judged as worse). A potential concern in the design of this study is that the 2 treatment arms be well matched in terms of extent of disease. At first this might seem problematic in that several patients in the surgery group had benign lesions such as bulky hyperkeratosis, hyperplasia, verruca, and so forth. In terms of the methods used in this study, inclusion of these patients poses no problem, however, in that the extent of removal for these lesions was of equivalent magnitude as that for the malignant lesions, coming near or to the ligament as the deep margin. For example, the extent of surgery for a large, bulky verruca involving the entire vocal fold was far greater than would be found in a patient with a small and localized invasive carcinoma. Another way of saying this by analogy is that the voice result for the same vocal fold wound (depth and extent of excision) would be the same in twin larynges, even if one wound resulted from removal of an early cancer and the other followed removal of normal mucosa. Note also that the voice results in the staging analysis (Fig 7) were the same for benign, Tis, and T1a lesions. It can also be noted that a greater number in the radiotherapy group had bilateral disease than did those in the surgery group. Two comments can help to alleviate this objection. As most laryngologists will agree, patients with T1b tumors often have diffuse, multifocal, and extremely superficial lesions as compared with those with unilateral lesions. Often some degree of hypomobility or even fixation is seen in initially unilateral tumors that have crossed the anterior commissure to become bilateral. In fact, in our T1b group, contralateral lesions were sometimes not even seen videostroboscopically, but only at the time of microlaryngoscopy, when the plan for superficial cordectomy was usually changed to radiotherapy on the basis of this finding. A second observation is that after radiotherapy the voices
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
of those with bilateral disease improved to the same degree as those with unilateral lesions, again supporting the superficiality of lesions. Finally, voice results after radiotherapy were good, and therefore the surgery group had to reach this same high standard to be considered the same. Long-held beliefs in the medical community can be hard to change. Such may be the case concerning this subject, the current study notwithstanding, because some surgeons and radiotherapists have only witnessed the weak, air-wasting vocal quality of persons who have undergone complete cordectomy16 rather than superficial cordectomy with narrow margins, as described here. Use of wider margins in the past may therefore explain why, in contrast to this study, previous reports have suggested better voice results after radiotherapy.22 Older concepts of cordectomy demand a much bigger operation (with greater degradation of the voice) than is required for cure. Although not yet standard surgical practice everywhere, the excision of superficial carcinomas has become more conservative in some places as greater awareness of the layer structure and microarchitecture of the true vocal fold has led to smaller margins overall, and to a deep margin that follows the contour of the fold, rather than cutting straight across it. This approach gives better voice results, as is clearly shown by us in the subgroup analysis for the different depths of excision and also by others.20 From an oncologic point of view, some may object on the grounds that use of such narrow margins as depicted in the figures has not been validated. Although oncologic validity was not our primary interest because the oncologic competitiveness of cordectomy even with smaller margins, if not the truly narrow margins described here, has been well supported by others,8,9 it can be noted that none of the surgical group reported here has experienced a recurrent cancer after a minimum follow-up of 3 years, for a cure rate of 100%, a rate also achieved by others.23 Recent social, economic, and political developments have mandated greater interest in the cost of various treatments. At Loyola University Medical Center, the cost of narrow-margin laser cordectomy in the outpatient setting (our routine practice) is less than 30% of that for full-course radiotherapy. Others have reached similar conclusions.24 Until now, the issues of cure rate, convenience, and cost lend narrow-margin laser cordectomy a distinct advantage over radiotherapy. The results of this study suggest that voice results also support the choice of narrow-margin superficial laser cordectomy as preferred to
DELSUPEHE et al
533
radiotherapy for lesions that can be removed while leaving most of the thyroarytenoid muscle intact. REFERENCES 1. Kanonier G, Rainer T, Fritsch E, et al. Radiotherapy in early glottic carcinoma. Ann Otol Rhinol Laryngol 1996;105:759-63. 2. Pellitteri PK, Kennedy TL, Vrabec DP, et al. Radiotherapy. The mainstay in the treatment of early glottic carcinoma. Arch Otolaryngol Head Neck Surg 1991;117:297-301. 3. Cragle SP, Brandenburg JH. Laser cordectomy or radiotherapy: cure rates, communication, and cost. Otolaryngol Head Neck Surg 1993;108:648-54. 4. Davis RK, Kelly SM, Parkin JL, et al. Selective management of early glottic cancer. Laryngoscope 1990;100:1306-9. 5. Hirano M. Structure of the vocal fold in normal and disease states: anatomical and physical studies. In: Proceedings of the Conference on the Assessment of Vocal Pathology. Rockville (MD): American Speech-Language-Hearing Association; 1981. ASHA Reports 11. 6. Kahane JC. Histologic structure and properties of the human vocal folds. Ear Nose Throat J 1988;67:322-30. 7. Hirano M, Kurita S, Nakashima T. The structure of the vocal folds. In: Stevens KN, Hirano M, editors. Vocal fold physiology. Tokyo: University of Tokyo; 1981. p. 33. 8. Hirano M, Hirade Y, Kawasaki H. Vocal function following carbon dioxide laser surgery for glottic carcinoma. Ann Otol Rhinol Laryngol 1985;94:232-5. 9. Ossoff RH, Sisson GA, Shapshay SM. Endoscopic management of selected early vocal cord carcinoma. Ann Otol Rhinol Laryngol 1985;94:560-4. 10. McGuirt WF, Blalock MA, Koufman JA, et al. Voice analysis of patients with endoscopically treated early laryngeal carcinoma. Ann Otol Rhinol Laryngol 1992;101:142-6. 11. Blakeslee D, Vaughan CW, Shapshay SM, et al. Excisional biopsy in the selective management of T1 glottic cancer: a three year follow-up study. Laryngoscope 1984;94:488-94. 12. Perry A. [Letter]. Ann Otol Rhinol Laryngol 1994;103:85. 13. Epstein BE, Lee DJ. [Letter]. Cancer 1990;40:366-7. 14. Marks R, Clark K, Wetmore SJ. T1 squamous cell carcinoma of the true vocal cord [controversies]. Head Neck Surg 1988;(Suppl 1):S69-71. 15. Kleinsasser O. Tumors of the larynx and hypopharynx. New York: Thieme Medical Publishers; 1988. p. 159. 16. Cinberg JZ, Silver CE. Treatment of larynx tumors with the CO2 laser. In: Silver CE. Surgery for cancer of the larynx and related structures. New York: Churchill Livingstone; 1981. p. 154. 17. Fitch JL. Consistency of fundamental frequency and perturbation in repeated phonations of sustained vowels, reading, and connected speech. J Speech Hear Disord 1990;55:360-3. 18. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. 19. Altman DG. Practical statistics for medical research. London: Chapman & Hall; 1991. p. 403-9. 20. Kaufman JA. The endoscopic management of early squamous carcinoma of the vocal cord with carbon dioxide surgical laser: clinical experience and a proposed classification. Otolaryngol Head Neck Surg 1986;95:531-7. 21. Stoicheff ML, Ciampi A, Passi JE, et al. The irradiated larynx and voice: a perceptual study. J Speech Hearing Res 1983;26:482-5. 22. Benninger MS, Gillen J, Thieme P, et al. Factors associated with recurrence and voice quality following radiation therapy for T1 and T2 glottic carcinomas. Laryngoscope 1994;104:294-8. 23. Strong MS. Laser excision of carcinoma of the larynx. Laryngoscope 1975;85:1286-9. 24. Meyers EN, Wagner RL, Johnson JT. Microlaryngoscopic surgery for T1 glottic lesions: a cost-effective option. Ann Otol Rhinol Laryngol 1994;103:28-30.
Chicago, Illinois, and Leuven, Belgium
Voice quality remains the issue often used to support preference for radiotherapy in treatment of early glottic cancer. We therefore conducted a perceptual voice study in 2 groups, one treated with radiotherapy for malignant disease and the other with narrow-margin laser cordectomy for either malignant or extensive benign lesions. Sequential patients, 12 treated with radiotherapy and 30 with CO2 laser excision, were included. Voice samples were recorded before and at intervals after surgery. Ratings of validated judges were used for statistical analysis of various voice characteristics at each time point. Voice deteriorated temporarily after surgery as compared with the radiated group; however, at 6 and 24 months no significant differences were found between the groups. Preferential use of narrow-margin laser cordectomy for appropriate early glottic tumors can be supported not only for oncologic reasons but also on the basis of voice results, cost, and efficiency considerations. (Otolaryngol Head Neck Surg 1999;121:528-33.)
T
he current standard of care for early glottic tumors remains controversial. The 2 competing treatment options, radiotherapy and laser excision, each have reported cure rates of approximately 90%.1-4 Despite similar cure rates, in most places radiotherapy seems to be the preferred option, primarily because of concern that voice quality is poor after cordectomy. Radiotherapy is thought to give better voice results, although no long-term comparative results to detect late deterio-
ration of the voice after radiotherapy have been published.2 In recent years information concerning the microarchitecture and layer structure of the vocal folds has become widely available.5-7 This information has led the senior author (R.W.B.) to use narrow margins, not only at the mucosal periphery of early vocal fold carcinomas but also at the deep plane, because the vocal ligament has been found to be an excellent barrier to spread for many of these tumors. From the beginning of a 10-year experience that began in 1987, it was noted that not only cure rates but also voice results were surprisingly good and satisfying to patients who had undergone this highly conservative kind of surgery. Because of this experience and because of the costeffectiveness and convenience of narrow-margin laser cordectomy as compared with radiotherapy, our first choice for treatment of early glottic carcinoma has become narrow-margin laser cordectomy. However, previous reports notwithstanding concerning the good8-10 or serviceable11 voice results that can be achieved after partial cordectomy, some clinicians continue to believe that unacceptable deterioration of voice quality results from narrow-margin laser cordectomy as compared with radiotherapy.12-14 This continuing belief is perhaps the primary reason that laser excision has not gained more advocates. Therefore this study was conceived to determine whether the surprising equivalence of voice results apparent to us informally would be validated when we performed a formal study to compare the voice results of groups treated with these 2 different modalities. METHODS AND MATERIAL Patient Population
From the Department of Otolaryngology, Loyola University (Drs Delsupehe and Bastian); and the Department of Otolaryngology, University Hospitals Leuven (Drs Delsupehe, Zink, and Lejaegere). Presented at the Annual Meeting of the American Academy of Otolaryngology–Head and Neck Surgery, San Francisco, CA, September 7-10, 1997. Reprint requests: Robert W. Bastian, MD, Department of Otolaryngology–Head and Neck Surgery, Loyola University Medical Center, 2160 S First Ave, Maywood, IL 60153. Copyright © 1999 by the American Academy of Otolaryngology– Head and Neck Surgery Foundation, Inc. 0194-5998/99/$8.00 + 0 23/1/101571 528
Forty-two consecutive patients treated at by a single physician (R.W.B.) at Loyola University Medical Center between January 1991 and March 1994 for known or presumed early stage cancers were studied. The patient and lesion characteristics are listed in Table 1. Treatment Modalities Surgery. Outpatient microlaryngoscopy with the microspot CO2 laser was used in every case. The surgical goal in each case was complete excision of the tumor with a narrow mar-
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
DELSUPEHE et al
Table 1. Patient and lesion characteristics
Age (y) Mean Range Sex (M/F) Smokers (current/ former/never) Drinkers (current/ former/never) Staging Benign Tis T1a T1c T2
Radiotherapy group
Surgery group
63.7 50-83 11/1 2/7/1
65.7 43-80 25/5 6/14/1
3/2/2
8/2/6
0 0 5 6 1
13 5 10 1 1
529
A Significance
NS NS NS
B
NS, Not significant.
gin of normal tissue. In contrast to some recommendations,9,11,15,16 mucosal margins were as little as the width of the laser beam and never more than 2 mm. The deep margin followed the contour of the vocal fold similar to Hirano’s5 description rather than cutting straight across it full thickness, as has been depicted elsewhere16 (Figs 1-4). Radiation. Patients were selected for radiotherapy because of medical considerations, anatomic limitations that prevented a clear view of the lesion during microlaryngoscopy, patient preference (as in a singer who was told that the traditional wisdom was that voice was better after radiotherapy than after surgery), and bilaterality. Radiotherapy was administered with an 8-MV linear accelerator to bilateral ports with field sizes ranging from 5 × 5 to 6 × 6 cm. Wedges were used to establish dose homogeneity. The average duration of radiotherapy was 6 weeks, with a total tumor dose ranging between 60 and 64 Gy. Voice Samples Voices were recorded before therapy, early after treatment, and again at 6 and 24 months. All recordings consisted of 2 sentences of the same phonetically balanced passage.17 Of a potential total of 168 samples (42 patients × 4 time points), 139 available recordings were randomized by a statistician. To test intraindividual variability, 30 duplicate recordings chosen randomly from among the 139 were scattered randomly within the samples, to create a total of 169 samples to be rated by the judges. Voice Characteristics Assessed All recordings were evaluated by 8 judges (4 trained and 4 untrained). The trained judges were all speech pathologists. The following auditory-perceptual voice characteristics were rated: overall voice quality, loudness, air wastage or breathi-
Fig 1. T1a carcinoma of the right vocal fold. A, Small lesion outlined with narrow margin. B, After excision, the vocal ligament was approached but not fully exposed.
ness, fluency, sense of effort, and appropriateness of fundamental frequency. The untrained judges evaluated only the first 3 of these characteristics, whereas the trained judges rated them all. Scoring of overall voice quality, loudness, and fluency were based on the following rating scale: 1, poor; 2, fair; 3, good; and 4, excellent. For air wastage and effort, the following scores were used: 1, severe; 2, moderate; and 3, none. Finally, fundamental frequency (F0) was rated as follows: 0, too high or too low; and 1, acceptable. Statistics To determine whether both treatment groups were similar, we compared patient factors and lesion characteristics for both groups using the Fisher exact test. Interjudge variability was assessed for each voice characteristic at each of the four time points, using the Spearman correlation coefficient. Intrajudge variability was analyzed using κ statistics. For all measurements the weighted κ value was used. The data set for all 8 judges was composed of 6075 judgments. On the basis of interjudge and intrajudge variability results for all 8 judges, perceptual ratings of the 2 most consistent judges (a set of 2028 ratings) were used for further statistical analysis. Ratings of these 2 judges for each of the voice char-
530
Otolaryngology– Head and Neck Surgery November 1999
DELSUPEHE et al
A
B
C
D
Fig 2. T1a carcinoma of the right vocal fold. A, Arrows show approximate margin that will be used for this well-circumscribed lesion. B, A narrow mucosal margin has been made with the laser. C, The deep plane of excision is the vocal ligament, which is clearly exposed. The deep surface of the flap containing the lesion is clean. D, Resulting narrow-margin laser cordectomy defect. Note that excision did not transect the fold but followed the plane of the vocal ligament.
acteristics were compared between treatment modalities by use of the Mann-Whitney U test to seek statistical differences. The Wilcoxon test was used to compare the scores within 1 treatment group at different time points. In addition, both tests were used for subgroup analysis (ie, voice result compared with depth of surgical excision and tumor stage). The graphs were plotted with the mean scores of both judges for each item. In those few instances when the weighted κ value for an individual judgment (of a particular voice characteristic) showed only poor or fair strength of agreement (weighted κ < 0.5), this particular rating was omitted when the graph was plotted.18 RESULTS
Forty-two patients entered the study. Twelve patients underwent radiotherapy, and 30 underwent narrow-margin laser cordectomy. Both groups were comparable with regards to age, sex, alcohol and smoking habits, and preoperative staging (Table 1). In the surgery group 13 patients with benign lesions and 5 with carcinoma in situ were included. These patients were kept in the study because their lesions were extensive and the exci-
sions were true narrow-margin laser cordectomies of magnitudes equal to and sometimes greater than those of many of the carcinomas. All others were invasive carcinomas or carcinoma in situ. Interjudge variability for all 8 judges was mostly very good (r > 0.5).19 Overall, trained judges 2 and 3 correlated best with the other judges (highest r values). Intrajudge variability (weighted κ value) assessment again showed trained judges 2 and 3 to have the best correlations (mean weighted κ values of 0.7 and 0.92, respectively). Before treatment there was no significant difference between the 2 treatment modalities in the ratings of any of the voice characteristics. In other words, before treatment the 2 groups were indistinguishable with respect to not only overall quality but also air wastage, loudness, fluency, effort, and F0. There was an improvement over time in the voice quality of both groups. This improvement reached statistical significance in the surgery group for overall quality (P = 0.03) and F0 within 6 months, and for air wastage (P = 0.029), fluency (P = 0.04), loudness (P = 0.012) by 24 months.
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
A
B
DELSUPEHE et al
531
A
B
Fig 3. Larger T1a cancer of the left vocal fold. A, Note the bulk and convexity of the free margin. B, After excision, which in this case took mucosa, ligament, and a small amount of muscle, leaving muscle exposed (arrow).
In the radiotherapy group there was a statistically significant improvement for air wastage only at 6 months (P = 0.017). In the surgery group there was a temporary worsening of all voice characteristics early after surgery. This turned into a net improvement by 6 months after surgery that was sustained at 24 months. When voice results for both treatment modalities were compared over time (Fig 5), the curves were very similar. Evaluation of all of the voice parameters showed no statistical differences between the radiotherapy group and the surgery group 6 and 24 months after treatment with the Mann-Whitney U test. Subgroup analysis was done for the depth of excision in the surgery group (Fig 6). In group A, a plane within the lamina propria or the curved surface of the ligament was the deep margin (only the mucosa and part or all of the superficial layer of the lamina propria were removed). In group B the muscle was the deep margin (mucosa and ligament removed). In group C the superficial cordectomy included part of the thyroarytenoid muscle.8,20 For all of the voice characteristics there was a clear trend in favor of both groups A and B
Fig 4. T1a lesion of the right vocal fold. A, Note that the tumor crowds but does not cross the anterior commissure (white arrow). It lies against the anterior left vocal fold (black arrows). B, After excision, the internal perichondrium of the thyroid cartilage is visible (large arrow). Although full-thickness muscle is gone in a small area anteriorly, most of the muscle of the fold as a whole remains intact (small arrows).
compared with the deeper excision (group C). Because of the small sample size, this reached significance only for overall voice quality at 6 months (group A vs group C; P = 0.024) (Fig 5). In a second subgroup analysis we compared the voices according to the different stages of the lesions in both treatment groups (Fig 7). The preoperative staging correlated very well with the pretreatment voice quality, which tends to validate the judges’ ratings (larger tumors tended to have worse voices). After treatment in the radiotherapy group, the voices improved irrespective of staging differences. In the surgery group there is a trend toward worse voice quality after removal of bigger tumors. Figure 7 shows an example of the effect of stage on overall voice quality.
532
DELSUPEHE et al
Fig 5. Evolution of overall voice quality at 4 time periods: before treatment, early after treatment, and at 6 and 24 months after surgery or radiotherapy.
Fig 6. Evolution of overall voice quality, surgery group only, according to depth of excision.
DISCUSSION
This study evaluates various voice characteristics before and after treatment of early glottic tumors with either radiation therapy or narrow-margin laser cordectomy. Despite the current interest in so-called “objective” measures of vocal output (eg, aerodynamic and acoustic analysis) the use of only auditory perception21 of validated judges for this study was intentional. This is because we were specifically interested in the “real life” question: Can trained and untrained judges distinguish between treatment groups by listening carefully to the voice result? The basic answer is that even the best 2 of 8 judges were unable to do so once surgical healing was complete. Several findings tend to validate the methods and results achieved here. First, with respect to the judges, there was a high degree of interjudge and intrajudge consistency. Second, the voice characteristics of the 2 groups were judged to be the same before treatment to a statistically significant degree, suggesting that they were comparable groups in terms of pretreatment phonatory disturbance. Third, there was a clear and similar trend of voice improvement for most of the characteristics over time. Finally, results in the subgroup
Otolaryngology– Head and Neck Surgery November 1999
Fig 7. Evolution of voice quality in the surgery group according to substage.
analysis were exactly as would be expected (eg, voice results for larger excisions tended to be judged as worse). A potential concern in the design of this study is that the 2 treatment arms be well matched in terms of extent of disease. At first this might seem problematic in that several patients in the surgery group had benign lesions such as bulky hyperkeratosis, hyperplasia, verruca, and so forth. In terms of the methods used in this study, inclusion of these patients poses no problem, however, in that the extent of removal for these lesions was of equivalent magnitude as that for the malignant lesions, coming near or to the ligament as the deep margin. For example, the extent of surgery for a large, bulky verruca involving the entire vocal fold was far greater than would be found in a patient with a small and localized invasive carcinoma. Another way of saying this by analogy is that the voice result for the same vocal fold wound (depth and extent of excision) would be the same in twin larynges, even if one wound resulted from removal of an early cancer and the other followed removal of normal mucosa. Note also that the voice results in the staging analysis (Fig 7) were the same for benign, Tis, and T1a lesions. It can also be noted that a greater number in the radiotherapy group had bilateral disease than did those in the surgery group. Two comments can help to alleviate this objection. As most laryngologists will agree, patients with T1b tumors often have diffuse, multifocal, and extremely superficial lesions as compared with those with unilateral lesions. Often some degree of hypomobility or even fixation is seen in initially unilateral tumors that have crossed the anterior commissure to become bilateral. In fact, in our T1b group, contralateral lesions were sometimes not even seen videostroboscopically, but only at the time of microlaryngoscopy, when the plan for superficial cordectomy was usually changed to radiotherapy on the basis of this finding. A second observation is that after radiotherapy the voices
Otolaryngology– Head and Neck Surgery Volume 121 Number 5
of those with bilateral disease improved to the same degree as those with unilateral lesions, again supporting the superficiality of lesions. Finally, voice results after radiotherapy were good, and therefore the surgery group had to reach this same high standard to be considered the same. Long-held beliefs in the medical community can be hard to change. Such may be the case concerning this subject, the current study notwithstanding, because some surgeons and radiotherapists have only witnessed the weak, air-wasting vocal quality of persons who have undergone complete cordectomy16 rather than superficial cordectomy with narrow margins, as described here. Use of wider margins in the past may therefore explain why, in contrast to this study, previous reports have suggested better voice results after radiotherapy.22 Older concepts of cordectomy demand a much bigger operation (with greater degradation of the voice) than is required for cure. Although not yet standard surgical practice everywhere, the excision of superficial carcinomas has become more conservative in some places as greater awareness of the layer structure and microarchitecture of the true vocal fold has led to smaller margins overall, and to a deep margin that follows the contour of the fold, rather than cutting straight across it. This approach gives better voice results, as is clearly shown by us in the subgroup analysis for the different depths of excision and also by others.20 From an oncologic point of view, some may object on the grounds that use of such narrow margins as depicted in the figures has not been validated. Although oncologic validity was not our primary interest because the oncologic competitiveness of cordectomy even with smaller margins, if not the truly narrow margins described here, has been well supported by others,8,9 it can be noted that none of the surgical group reported here has experienced a recurrent cancer after a minimum follow-up of 3 years, for a cure rate of 100%, a rate also achieved by others.23 Recent social, economic, and political developments have mandated greater interest in the cost of various treatments. At Loyola University Medical Center, the cost of narrow-margin laser cordectomy in the outpatient setting (our routine practice) is less than 30% of that for full-course radiotherapy. Others have reached similar conclusions.24 Until now, the issues of cure rate, convenience, and cost lend narrow-margin laser cordectomy a distinct advantage over radiotherapy. The results of this study suggest that voice results also support the choice of narrow-margin superficial laser cordectomy as preferred to
DELSUPEHE et al
533
radiotherapy for lesions that can be removed while leaving most of the thyroarytenoid muscle intact. REFERENCES 1. Kanonier G, Rainer T, Fritsch E, et al. Radiotherapy in early glottic carcinoma. Ann Otol Rhinol Laryngol 1996;105:759-63. 2. Pellitteri PK, Kennedy TL, Vrabec DP, et al. Radiotherapy. The mainstay in the treatment of early glottic carcinoma. Arch Otolaryngol Head Neck Surg 1991;117:297-301. 3. Cragle SP, Brandenburg JH. Laser cordectomy or radiotherapy: cure rates, communication, and cost. Otolaryngol Head Neck Surg 1993;108:648-54. 4. Davis RK, Kelly SM, Parkin JL, et al. Selective management of early glottic cancer. Laryngoscope 1990;100:1306-9. 5. Hirano M. Structure of the vocal fold in normal and disease states: anatomical and physical studies. In: Proceedings of the Conference on the Assessment of Vocal Pathology. Rockville (MD): American Speech-Language-Hearing Association; 1981. ASHA Reports 11. 6. Kahane JC. Histologic structure and properties of the human vocal folds. Ear Nose Throat J 1988;67:322-30. 7. Hirano M, Kurita S, Nakashima T. The structure of the vocal folds. In: Stevens KN, Hirano M, editors. Vocal fold physiology. Tokyo: University of Tokyo; 1981. p. 33. 8. Hirano M, Hirade Y, Kawasaki H. Vocal function following carbon dioxide laser surgery for glottic carcinoma. Ann Otol Rhinol Laryngol 1985;94:232-5. 9. Ossoff RH, Sisson GA, Shapshay SM. Endoscopic management of selected early vocal cord carcinoma. Ann Otol Rhinol Laryngol 1985;94:560-4. 10. McGuirt WF, Blalock MA, Koufman JA, et al. Voice analysis of patients with endoscopically treated early laryngeal carcinoma. Ann Otol Rhinol Laryngol 1992;101:142-6. 11. Blakeslee D, Vaughan CW, Shapshay SM, et al. Excisional biopsy in the selective management of T1 glottic cancer: a three year follow-up study. Laryngoscope 1984;94:488-94. 12. Perry A. [Letter]. Ann Otol Rhinol Laryngol 1994;103:85. 13. Epstein BE, Lee DJ. [Letter]. Cancer 1990;40:366-7. 14. Marks R, Clark K, Wetmore SJ. T1 squamous cell carcinoma of the true vocal cord [controversies]. Head Neck Surg 1988;(Suppl 1):S69-71. 15. Kleinsasser O. Tumors of the larynx and hypopharynx. New York: Thieme Medical Publishers; 1988. p. 159. 16. Cinberg JZ, Silver CE. Treatment of larynx tumors with the CO2 laser. In: Silver CE. Surgery for cancer of the larynx and related structures. New York: Churchill Livingstone; 1981. p. 154. 17. Fitch JL. Consistency of fundamental frequency and perturbation in repeated phonations of sustained vowels, reading, and connected speech. J Speech Hear Disord 1990;55:360-3. 18. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. 19. Altman DG. Practical statistics for medical research. London: Chapman & Hall; 1991. p. 403-9. 20. Kaufman JA. The endoscopic management of early squamous carcinoma of the vocal cord with carbon dioxide surgical laser: clinical experience and a proposed classification. Otolaryngol Head Neck Surg 1986;95:531-7. 21. Stoicheff ML, Ciampi A, Passi JE, et al. The irradiated larynx and voice: a perceptual study. J Speech Hearing Res 1983;26:482-5. 22. Benninger MS, Gillen J, Thieme P, et al. Factors associated with recurrence and voice quality following radiation therapy for T1 and T2 glottic carcinomas. Laryngoscope 1994;104:294-8. 23. Strong MS. Laser excision of carcinoma of the larynx. Laryngoscope 1975;85:1286-9. 24. Meyers EN, Wagner RL, Johnson JT. Microlaryngoscopic surgery for T1 glottic lesions: a cost-effective option. Ann Otol Rhinol Laryngol 1994;103:28-30.