TL;DR
This study investigates how concurrent chemoradiotherapy (CCRT) affects vocal function in advanced head and neck cancer patients. By analyzing acoustic features of sustained vowels before and after treatment, the researchers found that both the tumor itself and the treatment significantly impact voice quality, with effects varying by tumor location. This highlights the need for location-specific assessments of organ function post-treatment.
What Problem Does This Paper Address?
Advanced head and neck cancers are often treated with concomitant chemoradiotherapy (CCRT), a combination of chemotherapy and radiation. While CCRT improves treatment outcomes like locoregional control and survival, it can also severely impact patients' quality of life, particularly their ability to swallow, eat, and speak. Vocal function, specifically, can be compromised by the tumor's presence (e.g., edema, obstruction) and the treatment's side effects (e.g., fibrosis, vocal fold atrophy, mucositis). Existing research often focuses on survival and major functional outcomes like swallowing, with less detailed investigation into the specific, nuanced effects of CCRT on voice quality. This paper aims to disentangle the contributions of the tumor versus the treatment to vocal dysfunction by analyzing voice acoustics before and after CCRT, considering different tumor locations.
Methods
The study involved 47 patients with advanced squamous cell carcinoma of the oral cavity, oropharynx, hypopharynx, larynx, or nasopharynx. Voice recordings of sustained /a/ vowels were collected before and approximately 10 weeks after CCRT. Acoustic analysis was performed using the PRAAT software to assess various voice parameters, including:
- Irregularity: Jitter (cycle-to-cycle frequency variation), shimmer (cycle-to-cycle amplitude variation), and standard deviation of fundamental frequency (f0SD).
- Spectral Tilt: Measures like STdB (difference in spectral intensity between frequency bands) and H1A3 (ratio of the first harmonic to the third formant) were used to assess glottal adduction and vocal fold closure speed.
- Nasality: Assessed by H1A1 (difference between the first formant and first harmonic) and NAS (difference between filter bands).
- Noise Components: BEDdB (band energy difference), harmonics-to-noise ratio (HNR), and glottal-to-noise excitation ratio (GNE).
Patients were categorized by tumor site: laryngo/hypopharyngeal (L), nasopharyngeal (N), and oral cavity/oropharyngeal (O). Statistical analyses included Kruskal-Wallis tests to compare groups, Mann-Whitney U-tests for pairwise comparisons, and paired t-tests to assess changes from pre- to post-treatment.
Datasets
The study utilized voice recordings from 47 patients enrolled in the RADPLAT-protocol, a randomized clinical trial. This dataset represents a specific cohort of patients undergoing CCRT for advanced head and neck cancer. No new public benchmark datasets were introduced.
Key Insights and Ideas
- Pre-treatment voice reflects lifestyle: Before treatment, patients' daily alcohol consumption was strongly correlated with vocal instability (f0SD), and smoking correlated with glottal closure speed (H1A3). This suggests lifestyle factors significantly pre-condition vocal function.
- Tumor location matters: Pre-treatment voice quality differed significantly based on tumor site, particularly for nasality (NAS) and harmonics-to-noise ratio (HNR). Nasopharyngeal tumors, for instance, were associated with higher nasality, likely due to anatomical obstruction.
- Treatment-induced changes are site-specific: Post-treatment, while some voice measures worsened across all patients (e.g., HNR, BEDdB, spectral tilt), the degree and nature of these changes varied by tumor location. Laryngo/hypopharyngeal patients experienced significant worsening in measures like shimmer, jitter, and GNE, suggesting more severe damage to vocal fold function and glottal closure.
- Disentangling tumor vs. treatment effects: The study suggests that for some sites (e.g., oral/oropharyngeal), nasality decreased post-treatment, implying tumor reduction was a factor. Conversely, increased nasality in laryngo/hypopharyngeal patients post-treatment points towards treatment-induced changes.
- Acoustic analysis is a valuable tool: The study underscores the utility of objective acoustic analysis for evaluating organ function and differentiating between tumor-related and treatment-related effects in head and neck cancer patients.
Key Results
- Pre-treatment: Alcohol consumption correlated with f0SD (p<.001, r=.696). Smoking correlated with H1A3 (p=.005, r=.403). Tumor site significantly affected NAS (p=0.023) and marginally HNR (p=0.056). Laryngo/hypopharyngeal patients consumed more alcohol than oral/oropharyngeal (p=.030) and nasopharyngeal (p=.006) groups.
- Post-treatment: Across all tumor sites, HNR (p=.033), BEDdB (p=.018), STdB (p=.006), and H1A3 (p=.009) worsened significantly.
- Tumor Site Differences (Post-treatment):
- GNE: Significantly worse for laryngo/hypopharyngeal vs. oral/oropharyngeal (p=.015).
- HNR: Significantly worse for laryngo/hypopharyngeal vs. oral/oropharyngeal (p=.003).
- Jitter: Significantly worse for laryngo/hypopharyngeal vs. oral/oropharyngeal (p=.019).
- Shimmer: Significantly worse for laryngo/hypopharyngeal vs. oral/oropharyngeal (p=.001).
- Nasality (NAS): Improved significantly in nasopharyngeal patients (p=.023).
- Within-group changes (post-treatment): Laryngo/hypopharyngeal patients showed significant worsening in shimmer (p=.030), GNE (p=.015), and nasality (p=.043). Oral/oropharyngeal patients showed significant worsening in HNRSD (p=.026), BEDdB (p=.019), spectral tilt (p=.017), and H1A3 (p=.003).
Limitations and Future Work
- Limitations: The study acknowledges that preliminary evaluations of radiation doses are ongoing and will provide further insight into treatment effects. The sample size for the nasopharyngeal group (N=7) is relatively small, which might limit the generalizability of findings for that specific subgroup. The study focused on sustained vowels, and further analysis of running speech is planned.
- Future Work: The authors plan to analyze running speech to gain a more comprehensive understanding of vocal function. They also intend to correlate these voice findings with outcomes for swallowing and trismus, which were assessed via patient questionnaires and clinician findings in the same trial. Further investigation into the role of specific radiation doses and lymph node irradiation is also planned.
Summary updated