If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence and reprint requests to Camilla Slot Mehlum, Department of Oto-Rhino-Laryngology – Head & Neck Surgery and Audiology, Odense University Hospital, Denmark. J.B. Winsløws Vej 4, Indg. 84, 1. sal, 5000 Odense C, Denmark
This study investigates vocal outcome after cordectomy by transoral CO2 laser microsurgery (TLM-cordectomy) in patients with laryngeal intra-epithelial neoplasia (LIN) or non-neoplastic lesions (NNL), for improved individual patient advice and potential adjustment of national treatment strategy by which patients suspected to have glottic LIN or T1a cancer are offered TLM-cordectomy, without prior biopsy.
Study Design
Prospective, longitudinal, quasi-experimental time series.
Methods
Consecutively included patients (n = 155) with LIN (n = 84) or NNL (n = 71) who underwent voice assessments before and after TLM-cordectomy. The multi-dimensional voice assessment protocol comprised voice and speech range profiles, aerodynamics, acoustic analysis, self-evaluated voice handicap, and perceptual auditory voice ratings.
Results
Median follow-up time was 195 (range 50-1121) days for patients with LIN and 193 (range 69-1294) days for patients with NNL. Statistically significant changes, LIN: in voice handicap index (VHI) and breathiness after TLM-cordectomy. Statistically significant changes, NNL: voice range profile (voice range area, intensity range, and frequency range) and VHI after TLM-cordectomy. All group-wise changes were to less disordered voices. Previous smokers had the largest decreases in VHI and breathiness. Patients with baseline VHI scores >65 had smaller increases in VHI, however 13-19% of the patients had increases in VHI above the clinically relevant threshold after TLM-cordectomy.
Conclusion
Overall, TLM-cordectomy in patients with LIN and NNL improved vocal outcome and our study thus supports the current Danish treatment strategy and improves the basis for proper patient advice. Multi-dimensional voice assessment is suggested preoperatively and six-nine months postoperatively, with focus on individual vocal differences and voice demands.
Vocal fold carcinoma often develops through a multistage process where normal mucosa transforms into premalignant lesions with increasing malignant potential (laryngeal intraepithelial neoplasia; LIN) finally resulting in invasive malignant carcinoma.
Current presurgical non-invasive diagnostic techniques provide valuable information of the lesions, but is not always able to distinguish LIN or early glottic cancer from non-neoplastic lesions (NNL),
applied for the treatment of LIN and early glottic cancer. In Denmark different protocols were followed in different departments until 2012. A suspicious glottic lesion would usually entail a simple biopsy and depending on the histological diagnosis the patients were offered radiation for carcinoma, and laserevaporation, phonomicrosurgery, or a conservative treatment strategy for LIN and NNL. In 2012, a Danish national treatment strategy, the DANGLOT protocol, was introduced
and the DANGLOT database was established. This new strategy meant that patients with glottic lesions suspected for LIN or T1a cancer were offered transoral CO2 laser microsurgery (hereafter collectively abbreviated as TLM-cordectomy) as a one-stage surgical procedure, without prior biopsy. The TLM-cordectomy is classified into subtypes I-VI according to the degree of resection and surgical approach.
Cordectomy type I-III was applied in the DANGLOT protocol. Simple biopsy before TLM-cordectomy was discouraged as the subsequent scarring would perhaps necessitate more extensive resections.
Further, the strategy ensured complete resection of the lesion for histological evaluation and was expected to reduce the risk of relapse. However, concerns were that the new treatment strategy caused an increased risk of postoperative voice problems. Especially patients in whom LIN or cancer were not detected could be disadvantaged by following the new strategy, as less invasive treatment options might have been preferred, if the diagnosis of NNL was known preoperatively. Based on past experience, a significant proportion of patients were expected to be overtreated despite an extensive preoperative examination program. This was later confirmed when 3-year data from the DANGLOT database (n = 261 patients) showed that 30.3% had NNL, 36.4% had LIN and 33.3% T1a cancer.
The DANGLOT protocol thus involved mandatory voice assessment before and after surgery, and according to the agreed research plan related to the DANGLOT protocol, the voice outcome was to be examined when sufficient data was available.
The vocal outcome after TLM-cordectomy have been investigated extensively in patients with malignant lesions, eg T1a glottic carcinomas,
Quality of life and voice outcome of patients treated with transoral CO2 laser microsurgery for early glottic carcinoma (T1-T2): a 2-year follow-up study.
yet similar studies of patients with LIN and NNL are sparse. Likewise, comparisons of vocal outcome depending on the depth and extension of the cordectomy
To our knowledge, only few studies have investigated the vocal outcome after cordectomy for LIN and NNL patients. Reports of self-evaluated voice handicap are conflicting, as both decreased Voice Handicap Index (VHI) (VHI-10, n = 27),
The fifteen patients with LIN were not separated from the rest of the sample (n = 62) and therefore conclusions regarding patients with LIN are difficult to draw. Lim et al used TLM with various techniques
and reported no significant differences in jitter, shimmer, or dysphonia as rated using the GRBAS protocol (evaluation of Grade, Roughness, Breathiness, Asthenia, Strain
); however, they did find a difference in noise-to-harmonics ratio.
Thus, the potential vocal effects and side effects of TLM-cordectomy in patients with LIN and NNL have not been investigated in depth, and studies using reliable assessment or rating approaches have been recommended.
The objective of this study was to assess the vocal outcome after TLM-cordectomy for LIN and NNL with a multi-dimensional voice assessment protocol, with the aim of 1) potentially adjusting the national treatment strategy into a differentiated approach, and 2) provide patients with a more specific prognosis of the voice before planned TLM-cordectomy.
METHODS AND MATERIALS
The design of this study was prospective, longitudinal, quasi-experimental time series (self-controlled).
Patients and inclusion criteria
The patients for this study were recruited from the Danish DANGLOT database. Patients suspected for glottic LIN or T1a cancer were treated according to the DANGLOT protocol
and registered prospectively into the DANGLOT database in the five national tertiary centers of Head and Neck cancer between September 1, 2012 and December 31, 2018. Unfortunately, only three of five centers registered voice assessment data. Inclusion criteria for the DANGLOT protocol were: (1) performed TLM-cordectomy for suspected glottic LIN or T1a cancer as evaluated by stroboscopy and/or highspeed digital imaging and/or enhanced endoscopy (Narrow Band Imaging, NBI, Olympus®, or Storz Professional Image Enhancement System, SPIES, Karl Storz®), and (2) no prior biopsy or other vocal cord surgery within the last 12 months and not more than once ever at the vocal cord harboring the lesion. Patients treated according to the DANGLOT protocol with a histological diagnosis of LIN or NNL, who underwent preoperative and postoperative voice assessment, were included in this study. Voice data from patients diagnosed with T1a cancer are analyzed in another ongoing study. All TLM-cordectomies were performed by experienced laryngeal surgeons.
The World Health Organization (WHO) histological classification system from 2005
Voice parameters were analyzed groupwise for LIN patients as posthoc analysis did not reveal any significant differences between subgroups (data not shown elsewhere).
Ethics
The Danish Data Protection Agency and the Regional Committees on Health Research Ethics in Southern Denmark approved the study. The patients gave written informed consents prior to participation.
Data collection, voice assessment procedure, and data analyses
Voice assessments were planned before surgery and six months after surgery. The first postoperative voice assessment was set at 6-months because inflammation, localized edema, and consequently scarring, that might affect voice quality in the first postoperative months, are expected to have stabilized at that point.
Quality of life and voice outcome of patients treated with transoral CO2 laser microsurgery for early glottic carcinoma (T1-T2): a 2-year follow-up study.
Speech & language pathologists (SLP) and SLP students with special training in the specific voice assessment protocol collected the data in each center. The voice assessments followed the European Laryngological Society's (ELS) guidelines for voice assessments
A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the committee on phoniatrics of the European Laryngological Society (ELS).
with a few adjustments. The following analysis were made:
Voice range profiles
Complete voice range profiles (VRP) using the Voice Profiler® equipment replaced recordings of frequency ranges and softest intensities. The VRP assessment is clinically reliable
and recordings using this equipment can be conducted in standard quiet office surroundings. We measured frequency and intensity measures (minimum, maximum, and ranges) and VRP area as the dependent variables (Figure 1). Hallin et als
protocol was followed except that all vowels were included instead of the recommendation of using (α:) consistently. This multi-vowel approach was chosen as it was preferred in the participating clinics at that time.
FIGURE 1Example of baseline (top left) voice range profile and follow-up voice range profile (top right) from a female patient, and a voice range profile from a healthy person (lower mid). X-axis: semitones and fundamental frequencies; y-axis: decibel sound pressure level. The frequency range; highest and lowest frequency; intensity range; highest and lowest intensity; and VRP area is depicted. The color of each VRP cell denotes time used per cell.
Speech range profiles (SRP) assessment was added to achieve objective measures of habitual speaking fo and intensity. In the SRP mean fo measured in hertz (Hz) and mean sound pressure level (SPL) measured in decibel (dB) of the habitual speaking voice were the two dependent variables. The patients spoke for 60 seconds about their day, while the SLP was silent. If they made breaks longer than a few seconds, the instruction was repeated and the recording retaken. They were instructed to breathe and speak normally. If it was too difficult to sustain running speech for 60 seconds, the patients were instructed to describe the office surroundings as a supplement.
Aerodynamics
Maximum phonation time (MPT) and forced vital capacity (FVC) were obtained, and phonation quotient (PQ) calculated, both were considered dependent variables. The assessments were obtained under guidance and encouragement from the SLP. Maximum phonation time (MPT) was obtained by an inspiration followed by phonating as long as possible on the vowel (α:). The longest sample out of three was used in the analysis. In the FVC measurement, the patient was instructed to inhale deeply, then exhale as hard and long as possible into the spirometer until the next inhalation was inevitable. The highest FVC result out of three samples was used in the calculation of PQ. PQ was calculated with the following formula: (PQ) = Vital Capacity (mL)/MPT (s).
A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the committee on phoniatrics of the European Laryngological Society (ELS).
A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the committee on phoniatrics of the European Laryngological Society (ELS).
The Cepstral Peak Prominence (CPP) analysis can be performed as an alternative acoustic voice analysis, as recommended in the American Speech and Hearing Association's (ASHA) guidelines (2018)
Recommended protocols for instrumental assessment of voice: American speech-language-hearing association expert panel to develop a protocol for instrumental assessment of vocal function.
The CPP measures the relationship between periodic and aperiodic energy in sustained vowels (CPPv) or connected speech (CPPs); thus, it is a measure of dysphonia more generally. The CPP measures were included as dependent variables.
The voices in the present study were classified regarding signal type on a scale from I to IV by two SLPs, who visually judged spectrogram samples independently using Titze's signal typing system from 1994.
Titze IR. Workshop on voice acoustic analysis summary statement. Work shop on: Voice Acoustic Analysis Sponsored by the National Center for Voice and Speech; 1994b; February 17-18, 1994; Denver, CO
The signal typing was conducted to assess whether MDVP analysis could analyze the voice samples reliably, and thus the signal typing was not a part of the study results.
The SLPs were blinded to whether the samples belonged to patients or controls. They discussed the spectrograms in cases of disagreement. In a very few cases, where agreement could not be reached, the samples were left untyped. Data from these patients were still included, as the only purpose with the signal typing was to determine whether perturbation analysis or alternative analysis should be deployed.
In cases of disagreement, the sample was discussed to achieve consensus. In a very few cases, consensus could not be reached, and the sample was left unclassified yet still included in the study, as the main reason for signal typing was to examine, if the data material was suitable for perturbation analysis or not.
Self-evaluated voice handicap
The patients were instructed to complete the VHI (VHI-30) questionnaire.
Patients, who were unable to read or fill in the questions themselves, had the questions read aloud by the clinician. Only questions regarding the wording were answered. The minimal relevant difference from before to after was set conservatively at 13 points or more in order to detect small differences. This is elaborated in the Discussion section.
Perceptual auditory voice evaluation
Three SLPs experienced in perceptual auditory evaluation of voices, rated sound recordings of the patients’ text reading. The excerpts selected for evaluation were the middle part of the standard passage “The Wind and the Sun”. The Danish dysphonia assessment (“Protocol for auditory evaluation of dysphonic voices”) was deployed.
Auditiv stemmeanalyse—referencemateriale til brug ved auditiv beskrivelse af dysfoniske stemmer.[Auditory Voice Analysis—Reference Voice Sample Material for Use in Auditory Description of Dysphonic Voices].
Center for Communication and Rehabilitation, Roskilde and Institute of General and Applied Linguistics, University of Copenhagen,
Copenhagen, Denmark2003
Auditiv stemmeanalyse—referencemateriale til brug ved auditiv beskrivelse af dysfoniske stemmer.[Auditory Voice Analysis—Reference Voice Sample Material for Use in Auditory Description of Dysphonic Voices].
Center for Communication and Rehabilitation, Roskilde and Institute of General and Applied Linguistics, University of Copenhagen,
Copenhagen, Denmark2003
was conducted with consensus listening and pilot-ratings.
For the purpose of this study, roughness and gratings were aggregated into one variable and the term “Middle register” was omitted. This was done to adjust the protocol to the specific auditory features of this patient group. In addition, pitch and loudness ratings were excluded, as the fundamental frequency (fo) and SPL were measured in the SRP. Thus, the eight dependent variables were overall grade of dysphonia, hyperfunctionality, hypofunctionality, roughness and/or gratings, diplophonia, breathiness, aphonia, and glottal fry. A four-point scale (0-3) was employed, denoting “no,” “mild,” “middle” and “severe” grade of the parameter in question to match the standard reference sound file material. The SLPs knew the study purpose. They knew which recordings belonged to the same patients as well as the sex, but not if a sample was recorded before or after surgery or the histological diagnosis for the individual patients. They knew the patients’ gender and the study purpose. Due to time constraints, one SLP only rated one-third of the samples.
Equipment and facilities
All sound recordings were conducted in clinical office surroundings and followed the ASHA guidelines from 2018 for acoustic measurements.
Recommended protocols for instrumental assessment of voice: American speech-language-hearing association expert panel to develop a protocol for instrumental assessment of vocal function.
These comprise using a microphone preamplifier, a calibrated head-mounted microphone, and converting the signal to a digital format. C-weighting is preferred for sound level measuring to avoid uneven measurements across the frequency range, which might occur with A-weighting. Minimum sampling rate should be 44.1 kHz with a minimum resolution of 16 bits, and noise at least 10 dB lower than the minimum vocal sound level.
Recording distance for the measurements (vowel, reading a standard passage, VRP, and SRP) was 30 cm from microphone to lower lip. Distortion from nasal breath etc was avoided by using microphone windscreens.
Hillenbrands algorithms and SpeechTool system for the Windows PC operating system were used for the CPP analysis.
The second vowel, out of three sustained vowels, was chosen for the analysis, unless this clearly deviated from the two other vowels based on visual and auditory judgment.
For the recording of sustained vowel and text reading, Olympus LS-11 handheld battery-powered recorders were utilized. This device records without adding internal ventilation noise. The automatic sound level control was switched off and signals were stored digitally in .wav file format.
FVC was assessed by a handheld spirometer, type Micro Medical Limited MV or matching devices dependent on center. The Danish validated version of the VHI (VHI-30) questionnaire was used for self-evaluated voice handicap.
This model corrects for missing data based on the observed data. Assumption checks were made for acceptable levels of homoscedasticity and normality of the residuals when assessed by quantile-quantile normal plots. Further, Shapiro Wilks test was deployed to check for normality of the data. For variables, where random effects did not meet the assumption of normality of the residuals, the model was fitted with robust estimation. Fixed effects were time (baseline/follow-up), sex, resection type (I, II, III), age, and tobacco use (never/previous smoker, current smoker). Interaction terms were time and sex, and time and tobacco.
For ordinal data with less than ten measurement points, ie the auditory perceptual variables, ordered logistic regression was applied. Confounding factors were sex, tobacco use, and resection type. Cronbach's alpha (α)
was used for estimating internal consistency on the auditory perceptual data rated by all three SLPs with α values below 0.5 considered “insufficient” reliability, 0.5 ≤ α < 0.6 “poor” reliability, 0.6 ≤ α < 0.7 “questionable” reliability, 0.7 ≤ α < 0.8 “acceptable” reliability, 0.8 ≤ α < 0.9 “good” reliability and α 0.9 or higher as “excellent” reliability.
Scatterplots were used for depicting the differences between baseline and follow-up. By Pearson's correlation coefficient the correlation between total VHI score and the remaining variables was examined, and between VHI changes and the remaining variables. Correlation (r) (positive or negative) values below 0.30 was regarded “poor” correlation, 0.30 ≤ r < 0.50 as “moderate” correlation, 0.50 ≤ r < 0.70 as “good” correlation, and 0.70 or higher as “strong” correlation.
A priori power calculations were conducted on the dependent variables (α = 0.05, power 0.80) and the estimated sample sizes varied from a minimum of 36 to a minimum of 39 patients in each group to detect clinically relevant differences based on estimated clinically significant differences for this patient group. Examples of these were 6 dB in SPL range, 4 semitones (ST) in ST range, and 0.5 in CPPs. P-values of less than 0.05 were regarded statistically significant throughout the study.
RESULTS
Figure 2 provides an overview of the inclusion flow of patients. Out of the 196 patients who underwent baseline voice assessment, 155 patients were also assessed at follow-up and thus included in the study.
FIGURE 2Flowchart of patient inclusion process. Percentages are provided from the initially registered total of 196 patients (LIN, n = 102; NNL, n = 94). Abbreviations: LIN, laryngeal intraepithelial neoplasia; NNL, Non-neoplastic lesion; n, number.
Table 1 shows the characteristics for patients who underwent baseline and follow-up voice assessments. The number of patients, age distribution and smoking habits were comparable in the two groups of patients with LIN and NNL, respectively. The median follow-up time was 195 days for patients with LIN (range 50-1121) and 193 days for patients with NNL (range 69-1294). Six patients with LIN and five with NNL were assessed within the first five months after surgery. Eight patients with LIN and ten with NNL were assessed more than a year after surgery. These patients were included in the analysis, however, did not change the results in any direction.
TABLE 1Baseline Patient Characteristics and Follow-up Intervals for the 155 Patients Who Underwent Baseline Assessment, and One or Two Follow-ups (n [%])
12 patients with NNL diagnosis are registered as being disease-free after surgery for LIN.
(18%)
Potential recurrence
4 (5%)
1 (1%)
Disease-free after re-surgery for carcinoma
2 (2%)
0
Other
4 (5%)
3 (4%)
n/s
12 (14%)
12 (17%)
LIN diagnosis
Carcinoma in situ (LIN III)
6 (7%)
-
Severe dysplasia (LIN III)
24 (29%)
-
Moderate dysplasia (LIN II)
15 (18%)
-
Mild dysplasia (LIN I)
29 (35%)
-
Dysplasia degree n/s
10 (12%)
-
The table shows number of patients in numbers (n) and percent. SD, standard deviation; LIN, laryngeal intraepithelial neoplasia; NNL, non-neoplastic leukoplakia; AC, anterior commissure; n/s, not stated.
Significant differences (P < 0.05) are marked in bold.
† 12 patients with NNL diagnosis are registered as being disease-free after surgery for LIN.
‡ Time for registering last status can deviate from the time of voice assessment.
§ Ceased smoking >3 months before TLM-cordectomy.
║ Cigarettes per day for the current smokers. T tests used for comparing groups.
Table 2 shows the results of the voice assessments at baseline and postoperatively for the patients with LIN. There were statistically significant changes in means in three variables after surgery: total and physical VHI scores, and breathiness as rated by auditory perceptive assessment. The changes were to a less disordered voice.
TABLE 2Voice Assessments; Patients With LIN at Baseline and Follow-up (n = 84)
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
1.0 ± 0.1
1.0–1.1
–
–
–
Notes: Significant differences (P < 0.05) are marked in bold. Abbreviations: LIN, Laryngeal Intraeptithelial Neoplasia; SD, standard deviation; CI, confidence interval; diff, difference; VRP, voice range profile; fo, fundamental frequency; ST, semitones; Hz, hertz; m, males; f, females; dB, decibel; SRP, speech range profile; SPL, sound pressure level; MPT, maximum phonation time; sec, seconds; PQ, phonation quotient; l/sec, liters per second; CPP, cepstral peak prominence; OR, odds ratio; SE standard error.
Fewer assessments/only performed in one center.
† Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
The following effects of sex and smoking were shown for vocal outcome, with parenthesis showing mean coefficients and P-values. The females had an increase in min SPL of the VRP at the follow-up (+5.3; P = 0.002). Only the previous smokers had statistically significant decreases in total VHI and breathiness (VHI: previous smokers, -17, P = 0.047, non-smokers, P = 0.102; current smokers, P = 0.224, breathiness: previous smokers, odds ratio (OR) = 0.15, P = 0.022; non-smokers, P = 0.500; current smokers, P = 0.207).
NNL
The patients with NNL experienced more postoperative changes than the patients with LIN as shown in Table 3. As for the LIN group, all significant changes in means were to a less disordered voice. Statistically significant changes were found in three VRP-measures (larger SPL range and area, and higher frequency range (ST range)), fo mean for the females, and all four VHI scores.
TABLE 3Voice Assessments; Patients With NNL at Baseline and Follow-up (n = 71)
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Notes: Significant differences (P<0.05) are marked in bold. Abbreviations: NNL, Non-neoplastic Lesion, SD, standard deviation; CI, confidence interval; diff, difference; VRP, voice range profile; fo, fundamental frequency; ST, semitones; Hz, hertz; m, males; f, females; dB, decibel; SRP, speech range profile; SPL, sound pressure level; MPT, maximum phonation time; sec, seconds; PQ, phonation quotient; l/sec, liters per second; CPP, cepstral peak prominence; OR, odds ratio; SE standard error.
Fewer assessments/only performed in one center.
† Variable not analyzable; too few categories. The data were analyzed with linear mixed models and ordered logistic regression as described in the Statistics section.
Smoking status and sex had effects on some vocal outcomes. Current male smokers had a decrease in SPL range (-13.4, P = 0.007) as opposed to the mean group difference. Non-smokers had a decrease in CPPv at the follow-up (CPPv, -1.7; P = <001; previous smokers 0.069; current smokers 0.488). In the total VHI, the significant decreases were only found among the previous smokers (-17.4, P = 0.047; non-smokers, P = 0.102; current smokers P = 0.224).
Individual voice handicaps after TLM-cordectomy
The individual course of self-evaluated voice-problems is depicted in Figure 3. The lower horizontal lines in each scatter plot indicate “no change” in VHI. Patients figuring below this line had reduced VHI score postoperatively, and patients above the line had an increased VHI score after surgery.
FIGURE 3Scatterplots of individual changes in VHI. Scatterplots. LIN, (left); NNL (right). x-axis: baseline VHI total score; y-axis: difference between baseline and follow-up. Lower horizontal lines: no change between baseline and follow-up; top horizontal lines: threshold for clinically relevant increase (≥13 points)
.. Triangles: less voice handicap at follow-up than baseline, squares: small increase in voice handicap at follow-up (1–12 points). Circles: increase in voice handicap ≥13 points.
(horizontal top lines) from baseline to follow-up (LIN, 19%, n = 16; NNL, 13%, n = 9). These patients all had a baseline score of <65 points, which is slightly above half of the maximum obtainable VHI score. There were potential ceiling and floor effects; the total VHI score can be any number between 1 and 120, thus patients with baseline scores >107 are not able to reach a significant increase of 13 (and likewise at the other end of the scale). We did not detect a ceiling effect as no patients who approached maximum score at baseline had a large increase in VHI.
Pearson's correlation analysis was deployed to assess if there was a correlation between one or more measures at baseline and the changes in self-evaluated voice handicap on the total VHI score (VHI-change) from baseline to follow-up. The analysis showed strong (LIN, r = 0.71) and good (NNL, r = 0.53) correlations respectively, between the baseline VHI total score and VHI-change. The remaining measures had poor to moderate correlations (r < 0.50). As the histological diagnosis is unknown at baseline, the correlation analysis is more clinically applicable, if it includes all patients. In the compiled cohort, results were similar: “good” correlation (r = 0.63) between baseline VHI total score and r < 0.50 for the remaining measures. When separating the patients into previous smokers and current smokers, correlations were “strong” respectively “good” (previous smokers, r = 0.86; current smokers, r = 0.56). A separate correlation analysis between the baseline VHI total score and VHI-change for the non-smokers made no sense due to the low number of patients in this category.
Reliability of the auditory perceptual analysis
The levels of reliability between the three SPLs´ auditory perceptual ratings were determined by Cronbach's alpha analysis. Good reliability was found in aphonia (α = 0.87), hyperfunctionality (α = 0.81), acceptable reliability in roughness/gratings (α = 0.79), overall grade of dysphonia (α = 0.78), and breathiness (α = 0.75), and insufficient reliability in diplophonia (α = 0.46) and glottal fry (α = 0.32).
DISCUSSION
This study is the first to assess the vocal outcome after TLM-cordectomy in patients with LIN and NNL. The study used an extensive set of voice assessments with a protocol based on the ELS protocol, though modified to fit the diagnosis of LIN and NNL. The results indicate that TLM-cordectomy performed by experienced laryngeal surgeons does not harm the voices of patients with LIN or NNL when assessed on a groupwise basis. On average, the patients reported less voice handicap after TLM-cordectomy than preoperatively, which is in accordance with Koss et al,
The drop out percentage of 21% was to be expected in a clinical study in which patients often already knew that they were free of disease when the check-up was scheduled. Smoking is known to affect the voice negatively
The aim of this study was not to investigate the vocal differences between different status of smoking, however, we did compare differences in changes between patients who smoked at baseline and had ceased smoking >3 months before TLM-cordectomy (previous smokers), and those who had never smoked. Having ceased smoking at baseline had some positive effects on breathiness and VHI. Further potential side-effects from tobacco use remain undetected in this study as the low number of patients who had never smoked prevents comparison.
Despite the overall improved vocal outcomes, between 13% (NNL) and 19% (LIN) of the patients reported increases in VHI above the threshold for clinically relevant increases in their postoperative VHI after TLM-cordectomy. We chose a conservative difference in score of 13 points to be clinically relevant even though larger differences have also been suggested.
We found it important to detect quite small differences in order not to favor the treatment. Two important findings that might support a clinician in future estimations of voice prognosis for an individual patient were: (1) patients with baseline VHI scores in the upper half of the scale (65 points and above) can be reassured, that their voice handicap is not likely to increase after the TLM-cordectomy and (2) patients who have ceased smoking have superior vocal outcomes.
The mixed model included adjustment for resection type, however in our dataset it made little sense to evaluate on whether more extensive resections caused worse voice outcome. Firstly, type I cordectomy constituted the majority of cordectomies, next, surgery type was not stated on 57 patients, and third, data on recurrence and resurgery, which could be any type, was missing. These two elements rendered inferences from data uncertain. Therefore, the results cannot elucidate potential differences between vocal outcome after type I and II resections of LIN or NNL, which has mostly been investigated in malignant lesions where Type II probably are deployed more frequently.
It has been suggested, that leaving the anterior commissure (AC) and vocal fold muscle intact cause better outcomes, however only Type I and II resections were registered in the database and no inferences could be made regarding this statement on basis of our data material. No differences in voice outcome between patients with and without AC involvement after surgery, could be detected either.
We did not account for hearing status and it cannot be completely ruled out, that an unaided presbycusis in some of the patients could have affected the patients’ voice at baseline.
However, the hearing thresholds for these patients will probably not have deteriorated significantly within the time from baseline to follow-up. Further, pathological presbylaryngis is common among elderly voice patients from the mid-seventies and can affect vocal quality, MPT, fo mean and self-perceived voice handicap.
Ageing of the voice could have influenced the lack of voice improvements in the study, however only 11 patients with LIN and three patients with NNL were 75 years of age or older at the time of surgery. These small numbers are not likely to have affected the results significantly. Furthermore, alcohol intake could have influenced the results, yet probably causes less damage to the voice than smoking
and was not systematically recorded in the database.
The use of a multi-vowel VRP protocol is not likely to have influenced the results much, as the recommended (α:) was included in the multi-vowel approach. (α:) has been found mostly to entail greater contours in the VRP
; ie the use of more vowels does not cause more variation, rather they are unnecessary as they do not contribute much to the VRP. Further, the intra-examiner variation of the assessment when using the multi-vowel approach is below the clinically relevant differences.
The sample sizes met the requirements of the a priori power calculation, and thereby reduced the risk of type II errors. Overall, the study has high external validity and results are applicable on other patient populations with LIN and/or NNL, before and after TLM-cordectomy. It is important to stress though, that the TLM-cordectomies were conducted by experienced surgeons in high-volume departments of Head and Neck cancer centers and vocal outcomes might have been less favorable if conducted by less experienced surgeons.
The interval of minimum six months from surgery to the follow-up was chosen as the vocal fold epithelium most likely has healed during this time.
Quality of life and voice outcome of patients treated with transoral CO2 laser microsurgery for early glottic carcinoma (T1-T2): a 2-year follow-up study.
The six patients with LIN and five with NNL that were assessed within the first five months after surgery, comprised only 7% of the data and is not likely to have caused any significant modifications of the results.
The purpose of this study was not to examine whether the patients´ voices improved to a “normal” level. Such a comparison would require creating a normative control group material for this patient group with some special traits, eg the majority had a long-term history of tobacco consumption which probably affected the voice prior to the detection of the lesions. If the present patient population had been compared to a healthy standard population with a smaller fraction of smokers, this would most likely have caused some bias due to the standard population being vocally superior. A research study of the voices of patients with LIN and NNL compared to a matched control group is in preparation in our institutions.
The study was a longitudinal multi-center study conducted in real-life clinical settings, which increase the external validity for similar hospital settings. Reversely, the clinical setting also caused significant limitation ie missing data on recurrences, reoperations and change in smoking status, and wide timespans for voice assessment due to cancelled and postponed visits. According to registered data, most patients were disease-free without recurrence at the time of their last visit in the clinic, yet this was not always the same visit as the vocal follow-up. Therefore, the precise figures on these elements remain unclear. If for instance, patients with suspected recurrence could have been distinguished from disease-free patients, the postoperative voice assessments would reflect the vocal outcome and could be more directly related to the TLM-cordectomy and not extraneous factors. In cases of recurrence this would most likely have been discovered quickly in routine follow-up and an ELS-Type I cordectomy probably would have been employed. Taking this into consideration, the conclusions of the study can be regarded a worst-case scenario for vocal outcome and should be interpreted with these limitations in mind.
The vocal outcome results in this study do not necessitate revisions of the Danish national treatment strategy. However, it is important to keep individual differences in mind. Due to the individual differences among patients, and persisting voice disorders found at the follow-up, we strongly recommend a thorough post TLM-cordectomy voice assessment after six to nine months and optional later assessment depending on the clinical situation. This is to offer potential follow-up, SLP therapy, or other individual support. Patients who have ceased smoking clearly have better voice prognosis, and patients with very high VHI scores (in the upper half of the scale) more often report larger decreases in voice handicap postoperatively.
CONCLUSION
This prospective, longitudinal study has shown that overall, TLM-cordectomy, when conducted by experienced surgeons, improve mean vocal outcome in LIN and NNL patients when compared to baseline measurements. Only few patients report increased voice handicap and especially important is that the risk of long-term voice problems in potentially overtreated patients (NNL) is considered low. Our study thus supports the current Danish treatment strategy which is based on one stage TLM-cordectomy without prior biopsy for patients suspected for glottic LIN or T1a cancer. More precise preoperative information about the vocal outcome can be provided. However, individual differences, vocal demands, and the long-term vocal outcome must be assessed for each patient and special attention should be kept to patients with limited vocal problems before surgery (preoperative VHI scores in the lower half of the scale). Smoking cessation is recommended as well as one set of postoperative voice assessment after six to nine months is recommended. Future research on whether the multi-dimensional set of voice assessments can be simplified for patients with LIN and/or NNL is warranted. Conclusions must be interpreted with caution if applied on other patient cohorts.
FINANCIAL DISCLOSURE INFORMATION
None.
DECLARATION OF COMPETING INTEREST
None.
SOURCE OF FINANCIAL SUPPORT OR FUNDING
The study is part of a PhD project that has received private grants. None of the grant providers has specific interest in the outcomes of the study.
ACKNOWLEDGMENTS
We acknowledge all participating patients and clinicians, as well as Danielle Stone and Thea Nielsen for assistance with signal typing and CPP analysis. Sören Möller, OPEN statistics OUH is acknowledged for statistical help, Anne Bingen-Jakobsen for performing perceptual auditory ratings, and Rasmus Ferdinandsen for helping with editing and organizing the sound files.
REFERENCES
Rohde M
Grontved AM
Krogdahl A
et al.
Aggressive elimination of precancerous lesions of the vocal cords to avoid risk of cancer.
Quality of life and voice outcome of patients treated with transoral CO2 laser microsurgery for early glottic carcinoma (T1-T2): a 2-year follow-up study.
A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the committee on phoniatrics of the European Laryngological Society (ELS).
Recommended protocols for instrumental assessment of voice: American speech-language-hearing association expert panel to develop a protocol for instrumental assessment of vocal function.
Titze IR. Workshop on voice acoustic analysis summary statement. Work shop on: Voice Acoustic Analysis Sponsored by the National Center for Voice and Speech; 1994b; February 17-18, 1994; Denver, CO
Auditiv stemmeanalyse—referencemateriale til brug ved auditiv beskrivelse af dysfoniske stemmer.[Auditory Voice Analysis—Reference Voice Sample Material for Use in Auditory Description of Dysphonic Voices].
Center for Communication and Rehabilitation, Roskilde and Institute of General and Applied Linguistics, University of Copenhagen,
Copenhagen, Denmark2003
Preliminary data were presented orally at 12th Congress of the European Laryngological Society, QEII Centre, London GB, May 16 – 19, 2018, and at the annual meeting for the Danish Society of Otolaryngology, Head & Neck Surgery, Nyborg, Denmark, April 26, 2019. Baseline data is also included in another study with different scope, which is submitted simultaneously to J Voice.
Significant differences (P < 0.05) are marked in bold.
00129-1&id=gr1.jpg){kind=link}
00129-1&id=gr2.jpg){kind=link}
00129-1&id=gr3.jpg){kind=link}