Introduction
Patients with glottic carcinoma often present to the ENT surgeon with progressive hoarseness or dysphonia. Transoral laser microsurgery has become a standard treatment for patients with early glottic carcinoma, and its use is being expanded to those with later stages of disease.Reference Canis, Ihler, Martin, Wolff, Matthias and Steiner1, Reference Pfister, Spencer, Brizel, Burtness, Busse and Caudell2 Transoral laser microsurgery has advantages over open partial laryngectomy techniques, such as decreases in morbidity and surgical complications.Reference Egawa, Shimane, Akiyama, Ikenoya, Uzuki and Sugimoto3 A growing body of scientific literature is attempting to examine the survival outcomes of transoral laser microsurgery in glottic carcinoma.Reference Chiesa Estomba, Reinoso, Velasquez, Fernandez, Conde and Hidalgo4 However, few large studies have examined the longitudinal functional outcomes of transoral laser microsurgery.Reference Spielmann, Majumdar and Morton5–Reference Lee, Kim, Kim, Noh, Kim and Oh7
In order to evaluate patients for further voice modification procedures, one must first understand the optimal time to return of voice function after transoral laser microsurgery. Aaltonen and colleagues reported an improvement in voice hoarseness up to 12 months post-operatively after transoral laser microsurgery, using a visual analogue scale.Reference Aaltonen, Rautiainen, Sellman, Saarilahti, Mäkitie and Rihkanen8 Other studies using the Grade, Roughness, Breathiness, Asthenia, Strain (‘GRBAS’) scale have noted similar times to improvement.Reference Rydell, Schalén, Fex and Elner9, Reference Mendelsohn, Matar, Bachy, Lawson and Remacle10
Of specific importance is the characterisation of voice outcomes on patient's quality of life as a function of time. In terms of effect on quality of life, the Voice Handicap Index has been used to characterise the outcomes of transoral laser microsurgery. Lee and colleagues reviewed 57 patients who underwent transoral laser microsurgery for tumour (T) stage T1 glottic carcinoma over three- and six-month periods.Reference Lee, Kim, Kim, Noh, Kim and Oh7 They noted a deterioration of Voice Handicap Index scores in the early (three-month) post-operative period and subsequent improvement at six months. Mendelsohn and colleagues, who reported on a small group, also showed deterioration at four months and subsequent improvement at six months.Reference Mendelsohn, Matar, Bachy, Lawson and Remacle10 Most recently, a Taiwanese group using the Voice Handicap Index-30 noted a deterioration only at one month but not at three months.Reference Chu, Hsu, Lee, Fu, Wang and Kao11 Common limitations of these studies are short follow-up duration, small study populations and unclear rate of return of voice function. Here we propose a reliable model that would enable one to predict the time to recovery.
Patient selection for transoral laser microsurgery remains a challenge, as both radiation and transoral laser microsurgery are first-line treatments in early glottic carcinoma.Reference Pfister, Spencer, Brizel, Burtness, Busse and Caudell2 A recent comparative meta-analysis identified no significant differences in voice outcome for early glottic carcinoma between the two treatment methods.Reference Greulich, Parker, Lee, Merati and Misono12
Appropriate patient selection should be based on objective patient demographic data, such as gender and age, as well as local tumour advancement based on tumour–node–metastasis (TNM) classification. Of interest, studies assessing injection thyroplasty have reported significant anatomical differences between genders.Reference Mau and Courey13 Others have noted that this difference is not present at the level of quality of life, as determined by Voice Handicap Index score.Reference Sardesai, Merati, Hu and Birkent14 Age has been studied as a predictor of voice outcome in laser cordotomy and injection thyroplasty, with no differences between those aged over 65 and under 65 years.Reference Sardesai, Merati, Hu and Birkent14, Reference Vilaseca, Huerta, Blanch, Fernández-Planas, Jiménez and Bernal-Sprekelsen15 Surgical cordotomy involving the anterior commissure or para-epiglottic musculature has been shown to disrupt vocal function.Reference Oswal, Remacle, Jovanvic, Zeitels, Krespi and Hopper16 While generally correlated with extended cordotomy, the effect of tumour stage on voice handicap has not been clearly studied in the literature. Together, these factors help guide the ENT surgeon in terms of patient selection and counselling.
Thus, the objectives of our study were: (1) to determine and model the effect of post-operative time on the Voice Handicap Index outcome after transoral laser microsurgery for glottic carcinoma; and (2) to determine the effect of age, gender and stage on the Voice Handicap Index outcome after transoral laser microsurgery for glottic carcinoma.
Materials and methods
Patient data
A prospective cohort study was conducted. All patients referred to the Queen Elizabeth II Health Sciences Centre who were treated with transoral laser microsurgery for glottic carcinoma, between 2002 and 2014, were included in the study. Patients were staged using flexible endoscopy and pre-operative computed tomography imaging. All patients were presented at the multidisciplinary tumour board and were recommended surgical resection by transoral laser microsurgery.
Follow-up Voice Handicap Index scoring was scheduled at 3, 12 and 24 months following surgery. Data points up to one year prior to diagnosis were excluded if patients had local recurrence of their disease.
The procedures were conducted in accordance with the ethical standards of the committee on human experimentation. Ethics approval was completed through the Nova Scotia Health Authority Research Ethics Board (‘REB’ number: 1020643).
Quantification and statistics
Functional outcome was based on the Voice Handicap Index 10.Reference Rosen, Lee, Osborne, Zullo and Murry17 This 10-item self-reported screening questionnaire is graded on a scale of 0–40, with 0 reflecting no disability and 40 representing severe voice handicap.
Patient characteristics were obtained from a detailed chart review, and included: age, gender, primary affected vocal fold and TNM classification. Missing data were considered ‘missing at random’, as this was unrelated to the actual values of the missing data. For age analysis, patients were grouped by age according to 25th, 50th and 75th percentiles.
Age and stage comparison was completed by two-way analysis of variance (ANOVA) with Dunn's multiple comparisons to pre-operative Voice Handicap Index scores. Gender comparison of Voice Handicap Index was completed by two-way student's t-test, for each time interval. Voice Handicap Index comparison over time was completed using a one-way ANOVA with Dunn's multiple comparisons to pre-operative Voice Handicap Index. All graphing and statistical analysis was completed using GraphPad Prism software, version 6.0c (GraphPad Software, La Jolla, California, USA).
Mathematical model
Non-linear fitting was completed to derive voice improvement rates. The Voice Handicap Index score was assigned as a function output ‘y’, such that the data could be fitted to a generalised exponential equation: y = b*e (a*x) + c. In this equation, ‘c’ and ‘a’ are parameter estimates of pre-operative Voice Handicap Index and Voice Handicap Index improvement rate (months−1) respectively; ‘x’ represents the corresponding time (months) for each Voice Handicap Index score, and ‘b’ is the span of the values. Non-linear regression returned the estimated values for the Voice Handicap Index parameters of score, improvement rate and span.
We also evaluated the linear hypothesis, such that patients’ data were fitted to the generalised linear equation: y = a*x + b. In this equation, ‘a’ is the estimate of improvement rate (months−1) and ‘b’ corresponds to the pre-operative Voice Handicap Index score.
The linear and non-linear models were compared using the extra sum-of-squares F test to determine the most likely model. Each model was further analysed using the Akaike information criterion, to determine its probability to produce the given dataset. Using the non-linear model, the time for 95 per cent of maximal voice handicap recovery was calculated.
Results
Patient demographics
A total of 87 primary early-stage glottic carcinoma patients were recruited to the study. Compliance rates at individual time points were: 43 per cent at 3 months, 41 per cent at 12 months and 77 per cent at 24 months. The gender split was 81.6 per cent male and 18.4 per cent female.
Of the patients, the left side of the vocal fold was primarily affected in 42.5 per cent and the right side was primarily affected in 54.0 per cent; 3.4 per cent were bilaterally affected. The T classification revealed a distribution of: 17.2 per cent carcinoma in situ (CIS), 18.4 per cent T1a, 19.5 per cent T1b and 44.8 per cent T2 glottic carcinoma (Table 1). The rate of local recurrence at five years was 16 per cent. Five-year disease-free survival was 97 per cent.
Characteristics of patient population
SD = standard deviation; TNM = tumour–node–metastasis; CIS = carcinoma in situ
Age analysis
Patients were distributed equally into four groups by age distribution: less than 62 years, 62–69 years, 70–76 years, and more than 76 years. Patients aged 61 years or younger showed a statistically significant decrease in Voice Handicap Index when pre-operative scores were compared to scores at 12 months and more than 24 months post-operatively (p = 0.01, p = 0.02). Similarly, patients aged 62–69 years showed a significant reduction in scores at more than 24 months post-operatively (p = 0.04). No significant post-operative improvement was found for patients aged 70–76 years or those older than 76 years (Figure 1).
Comparison of Voice Handicap Index (VHI) outcome by grouped age. *Indicates significant difference (p < 0.05)
Gender analysis
No significant differences were observed between male and female gendered patients at the pre-operative screening, or at 3 months, 12 months or more than 24 months post-operatively (Figure 2).
Comparison of Voice Handicap Index (VHI) outcome by gender.
Stage analysis
Patients were grouped according to T classification as: CIS, T1a, T1b or T2 glottic carcinoma. There were no statistically significant reductions in Voice Handicap Index scores for the CIS and T1a groups in our population. However, patients with T1b or T2 glottic carcinoma showed a significant reduction at more than 24 months post-operatively (p = 0.005, p = 0.03) (Figure 3).
Comparison of Voice Handicap Index (VHI) outcome by tumour (T) stage. *Indicates significant difference (p < 0.05). CIS = carcinoma in situ
Separately, patients were grouped according to the resection of the anterior commissure. Our study noted no significant differences between patients who underwent anterior commissure resection and those with the anterior commissure unaffected (Figure 4).
Comparison of Voice Handicap Index (VHI) outcome by anterior commissure (AC) preservation.
Time analysis
Patients’ data were assessed pre-operatively, and at 3 months, 12 months and more than 24 months post-operatively, to examine the trend in Voice Handicap Index over time. The average scores were: 17.7 (confidence interval (CI) = 14.5–20.9) pre-operatively, 13.1 (CI = 10.0–16.3) at 3 months, 12.00 (CI = 9.3–14.7) at 12 months, and 11.9 (CI = 9.8–14.0) at more than 24 months. This corresponded to a decrease in Voice Handicap Index of 4.5 (p = 0.08) at 3 months, 5.6 (p = 0.02) at 12 months, and 5.7 (p = 0.006) at more than 24 months post-operatively (Figure 5).
Overall Voice Handicap Index (VHI) data by time. *Indicates significant difference (p < 0.05)
Mathematical modelling
We attempted to develop a mathematical model that best describes voice handicap at each time-point post-operatively. The modelling of health outcome data is important to summarise the relationships between two variables, and it provides the foundation for the extrapolation of outcomes. Critical in voice outcomes are the rate variables, which can be compared across populations to help guide treatment and expectations.
First, we wanted to compare the use of linear and non-linear regressions for our dataset (Figure 6). Using linear regression as the null hypothesis, we found a statistically preferred model by the extra sum-of-squares F test with non-linear one-phase decay regression analysis (p = 0.03). Subsequently, using a one-phase decay logarithmic equation, we were able to model the improvement in voice handicap by the relationship: y = 5.7*e (−0.52x) + 11.9. This gave an 81.0 per cent probability of producing the data by Akaike information criterion, compared with a 19.0 per cent probability by the linear model. Finally, we used the non-linear model to predict the time to 95 per cent of maximal improvement in voice handicap, which gave 5.5 months.
Linear versus non-linear predictive models of Voice Handicap Index (VHI) score over time. *Indicates significant difference (p < 0.05)
Discussion
Transoral laser microsurgery provides unique benefits compared to other treatment options given its low side-effect profile, while maintaining an excellent rate of local control. However, the effect of surgical resection on long-term voice-related quality of life has not been well established.Reference Spielmann, Majumdar and Morton5–Reference Lee, Kim, Kim, Noh, Kim and Oh7 The strengths of this study over previous studies are: a large cohort of patients with primary early glottic carcinoma (n = 87), the longitudinal analysis of voice-related quality of life (over 24 months), and a model of post-operative voice outcome.
With rising rates of laryngeal cancer in the female population, it is important to identify any potential gender difference in patients with glottic carcinoma.Reference Kachuri, De, Ellison and Semenciw18 Our patient population had an average male to female gender ratio of 4.4:1. This ratio is consistent with recent Canadian studies on laryngeal carcinoma prevalence rates that indicate a male to female prevalence of 6.2:1,Reference Kachuri, De, Ellison and Semenciw18, Reference Xie, Semenciw and Mery19 and the reported ratio in Cummings Otolaryngology – Head & Neck Surgery of 3.8:1.Reference Flint, Haughey, Lund, Niparko, Richardson and Robbins20
The distribution of left-sided and right-sided primary tumours was similar, at 42.5 per cent and 54.0 per cent respectively. By comparison, the distribution of T classification favoured T2 staged lesions (44.8 per cent), which may contribute to the improvement we observed post-operatively. Taken together, our study population reflects the average North American population with early glottic carcinoma.
Previous studies have reported that increasing age is not associated with increased post-operative mortality or complication rate in head and neck cancer patients treated with surgery.Reference Kim, Kim, Shim, Lee, Sung and Kim21 However, we found that patients younger than 70 years reported an improvement in quality of life related voice symptoms, while those 70 years and older did not have the same pre-operative to post-operative improvement (Figure 1). This finding may suggest that older patients experience fewer voice-related quality of life difficulties than their younger peers pre-operatively, related to the effect of voice on professional voice use or communication in the workplace. Alternatively, older patients may have lower voice expectations post-operatively, or the potential to compensate for the surgical defect that remains following transoral laser microsurgery. Others have reported similar findings in patients with head and neck cancer.Reference Veldhuis, Probst, Marek, Noack, Ural and Adamietz22, Reference Laraway, Lakshmiah, Lowe, Roe and Rogers23
Male gender predominance for laryngeal carcinoma has been decreasing over the past two decades.Reference Kachuri, De, Ellison and Semenciw18 Rising rates of glottic carcinoma have been attributed to an increase in female smokers.Reference Kachuri, De, Ellison and Semenciw18, Reference Markou, Christoforidou, Karasmanis, Tsiropoulos, Triaridis and Constantinidis24 This increase warrants further study to explore any potential differences in a historically male-dominated disease. Anatomical and social differences between male and female gendered patients have led to proposed differences in surgical outcome or quality of life measures.Reference Mau and Courey13, Reference Lee, Gibson and Hilari25 Our study found no differences in transoral laser microsurgery treated patients in terms of Voice Handicap Index outcomes before surgery or during post-operative follow up.
Disease stage analysis has become a crucial issue when discussing treatment options with patients (radiation vs surgery).Reference Mendenhall, Werning, Hinerman, Amdur and Villaret26, Reference Agrawal and Ha27 At our institution, patients classified with CIS up to those with T2 disease are treated primarily with transoral laser microsurgery. The staging difference between T1a and T1b is critical, as it represents advancement of the tumour into the region of the anterior commissure tendon (Broyles ligament) that acts as a barrier to further disease spread outside of the larynx.Reference Sadoughi28 Resection in this area may be complicated by the formation of an anterior glottic web, which may further disrupt voice function.Reference Sadoughi28 Interestingly, a significant improvement in Voice Handicap Index was observed for stage T1b and T2 glottic carcinomas, while no clear improvement was observed for CIS or T1a patients. One possibility is that T1b and T2 glottic carcinoma represent a greater burden of disease, leading to higher pre-operative voice handicap. Alternatively, voice-related quality of life may be influenced by patient overall satisfaction and expectations. Strong conclusions regarding whether voice effects are independent on stage remain limited by study power.
We also investigated post-operative time as it related to voice handicap. We found progressive improvement at 3 months, 12 months and more than 24 months post-operatively. We speculate that this relates to the time needed for complete remodelling of the glottic tissue after resection. While others have reported an initial deterioration of voice handicap at one to three months post-operatively, we did not find the same trend.Reference Lee, Kim, Kim, Noh, Kim and Oh7, Reference Mendelsohn, Matar, Bachy, Lawson and Remacle10, Reference Chu, Hsu, Lee, Fu, Wang and Kao11, Reference Delsupehe, Zink, Lejaegere and Bastian29 One explanation is that our study population has a high proportion of T2 cases, resulting in higher pre-operative Voice Handicap Index scores. Deterioration of voice may be blunted because of this high pre-operative handicap. Alternatively, our study may not have the temporal resolution to detect this temporary deterioration. Furthermore, other studies have not noted a decline at the three-month time interval.Reference Chu, Hsu, Lee, Fu, Wang and Kao11
The trend for mathematical modelling has been popularised in the field of facial reconstructive surgery.Reference Deuflhard, Weiser and Zachow30 Modelling patient outcomes post-operatively provides a numerical standard for rate and time outcomes. It is also useful for measuring individual progress, and more broadly to objectively compare outcomes in the literature. Our modelling of voice outcome data predicts that, on average, patients will develop a 5.7-point improvement in their Voice Handicap Index score. Furthermore, given the exponential decrease, we predict that patients will have progressive improvement of their voice, receiving on average 95 per cent of their total improvement in voice handicap at 5.5 months post-operatively. This information can be used to plan additional procedures post-operatively, such as injection thyroplasty, to further improve voice outcomes following surgery.Reference Maccarini, Stacchini, Salsi, Padovani, Pieri and Casolino31
This study is limited by its use of self-reported questionnaires, which may be subject to recall, repeat or response biases. Furthermore, the study population is recruited from a single institution and is subject to institutional bias. Finally, given the individualised post-operative recovery period, we did not control for many of the confounders that may affect recovery time.
• Prior studies on post-operative Voice Handicap Index had limited longitudinal data and small cohorts
• Patients with early glottic cancer will have voice improvement up to 24 months post-operatively
• Voice handicap improvement is non-linear in nature
• Younger patients and patients with larger tumours show greater improvements in post-operative Voice Handicap Index
Conclusion
This study has demonstrated that patients with glottic carcinoma treated with transoral laser microsurgery show significant improvement in post-operative quality of life Voice Handicap Index scores up to 24 months post-operatively. No differences in Voice Handicap Index scores were observed between male and female gender. Patients younger than 70 years of age had significant improvement in Voice Handicap Index, compared to older patients whose Voice Handicap Index improvement was not significant. Patients with stage T1b and T2 cancers showed a greater improvement compared to those with CIS and T1a disease. We believe these factors should be considered when selecting and counselling patients on their treatment options for early glottic cancer.
Coupled with robust oncological outcomes and significant cost savings, the functional outcomes described in this paper help to support our institution's overwhelming preference for transoral laser microsurgery in managing patients with early glottic cancer.Reference Melong, Rigby, Bullock, Hart, Trites and Taylor32–Reference McNeil, Wilke and Taylor35
Competing interests
None declared
