Secondary analyses of swallowing efficiency and safety outcomes following thyroidectomy versus thyroidectomy plus prophylactic central neck dissection

Abstract

Background

Swallowing complaints are common following total thyroidectomy, though an exact mechanism of patient-reported swallowing symptoms following thyroidectomy is currently lacking. This secondary, blinded analysis of data collected in a randomized, controlled clinical trial hypothesized that patients randomly assigned to the central neck dissection group would exhibit increased aspiration and pharyngeal residue on videofluoroscopic swallowing evaluation, and reduced patient-rated swallowing outcomes, as compared to patients randomized to thyroidectomy alone. We further hypothesized that blinded analysis would reveal worse swallowing function two-weeks post-surgery when compared to their pre-operative status to explain qualitative patient-reported dysphagia symptoms.

Methods

Thirty-two participants randomized to total thyroidectomy treatment with or without central neck dissection underwent pre- and post-surgical evaluation of swallowing outcomes, including videofluoroscopic Penetration/Aspiration Scale ratings, Normalized Residue Ratio Scale measures of valleculae and pyriform sinus residue, and EAT-10 patient-rated outcomes.

Results

No statistically significant differences were found post-surgery between randomized treatment groups for patient-rated EAT-10 scores (p = 0.2406), penetration/aspiration scale (p = 0.4465) or Normalized Residue Rating Scale scores for either vallecular or pyriform sinus sites. When group data were combined for analysis of differences between pre- and post-operative swallow performance, no statistically significant differences were found in patient-rated EAT-10 scores (p = 0.1374), penetration/aspiration scale (p = 0.7588) or Normalized Residue Rating Scale scores.

Conclusions

Measures of penetration/aspiration and pharyngeal residue failed to substantiate perceptions of post-operative dysphagia reported by patients undergoing total thyroidectomy with or without central neck dissection.

Trial registration

ClinicalTrials.gov Identifier NCT02138214.

Introduction

It is well-established that dysphagia symptoms may occur before and after total thyroidectomy [1]. Prior to total thyroidectomy, over half of patients describe dysphagia at least some of the time [2]. A systematic review investigating dysphagia after uncomplicated thyroidectomy showed that dysphagia symptoms resolved by one year following surgery but without an identified physiologic cause [3]. Patient-rated quality of life surveys in thyroid cancer survivors have revealed dysphagia to be associated with worse quality of life scores [4, 5].

Risk factors for dysphagia symptoms after thyroidectomy remain poorly understood. Patient age over 50 years was found to be independently predictive of dysphagia after thyroidectomy [6]. Increased risk for dysphagia was found among patients requiring a larger resection due to cancer [7]. Measures such as intraoperative nerve monitoring and minimally invasive surgical approaches intend to avoid negative outcomes associated with total thyroidectomy. Despite preventive measures, post-operative complications such as injury to superior (SLN) or recurrent (RLN) laryngeal branches of the vagus nerve may occur, resulting in sensory or motor impairments impacting swallowing function [8]. Even in the absence of nerve injury, dysphagia complaints have been prevalent among post-operative thyroidectomy patients [8–10]. While prophylactic central neck dissection (CND) to prevent locoregional recurrence in papillary thyroid cancer has remained controversial in its effectiveness, systematic review suggests equivalent risk for RLN injury with or without CND [11]. Detailed analyses of dysphagia following total thyroidectomy with and without CND are lacking.

Recent qualitative investigation of dysphagia conducted as part of a randomized, controlled clinical trial investigating oncologic benefit and functional risks associated with prophylactic central neck dissection in treatment of node negative papillary thyroid cancer reported that 80% of participants acknowledged dysphagia symptoms 2-weeks following total thyroidectomy, gradually reducing to 17% by 6 months [12]. When symptom rates were compared against clinical data collected in the same trial, only 8% of participants met clinical criteria for a follow-up swallowing evaluation after the 2-week study point. Trial results reported a lack of documented aspiration or pharyngeal residuals on videofluoroscopic swallow study in either randomized group [13]. The disparity between patient-reported symptoms and clinician-rated dysphagia raised questions regarding the source of perceived swallowing difficulty. This study aimed to perform a secondary analyses on the clinical trial swallow data subset to determine whether blinded ratings of videofluoroscopic imaging compared to patient-rated quality of life would reveal a similar divergence. We hypothesized that patients undergoing total thyroidectomy would show significant differences between pre-operative and two-week post-operative study points in penetration/aspiration scale scores, pharyngeal residue and patient-rated outcomes, with those randomized to undergo neck dissection showing significantly worse measures than those without neck dissection.

Methods

This study was designed to evaluate swallowing function in papillary thyroid cancer patients before and after total thyroidectomy with or without prophylactic central neck dissection. Patients were enrolled in a randomized clinical trial at the University of Wisconsin-Madison (RCT—Clinical Trial Registration Number NCT02138214) [13], and informed consent was obtained for each participant under the approval of the Institutional Review Board (IRB 2014 − 0391). Inclusion and exclusion criteria are listed in Table 1. Patients were randomized to one of two surgical groups, total thyroidectomy with no prophylactic central neck dissection (TT) or total thyroidectomy with prophylactic central neck dissection (CND), using computer generated randomization communicated to the surgeon in a sealed envelope. After randomization, subjects were excluded if intraoperative evaluation showed evidence of lymph node involvement or failed to confirm a diagnosis of cancer.

Table 1.

Clinical trial NCT02138214 inclusion and exclusion criteria

Inclusion Criteria	Exclusion Criteria
Pre-operative diagnosis or suspicion of papillary thyroid cancer	Largest papillary thyroid carcinoma < 1 cm in size on ultrasound
No pre-operative evidence of cervical lymph node metastases	Previous thyroid surgery
No evidence of distant metastases	Concurrent active malignancy of another type
Age 21–73 years	Age < 21 or > 73 years
Ability to read and write in English	Inability to give informed consent or lacks decision-making capacity
	T4 tumor
	Pre-existing vocal cord paralysis
	Chronic neurologic condition which affects voice or swallow
	Laryngeal pathology warranting intervention that could affect voice or swallow function
	Becoming pregnant before surgery or at any time while on study

Demographic data including age, sex, race, ethnicity, surgical history, and cancer history were collected from consented participants and are summarized in Table 2. All participants underwent videofluoroscopic swallow studies (VFSS) one day to six weeks before surgery and again within two weeks after surgery; consistencies evaluated were thin liquid, pudding, and solid across two bolus presentations each. VFSS recordings were viewed by two experienced speech-language pathologist raters who were blinded to participant and study timepoint. The Penetration-Aspiration Scale (PAS) was applied by each rater for assigning depth of penetration/aspiration and participant response to penetration/aspiration events [14]. Additionally, the Eating Assessment Tool-10 (EAT-10), a self-administered symptom-specific outcome instrument for dysphagia, was completed by patients at these same time points [15]. To measure pharyngeal residue following swallows, VFSS images were loaded into ImageJ digital analysis software (ImageJ, National Institutes of Health, Bethesda, MD). Using the Normalized Residue Ratio Scale (NRRS) protocol previously described [16], pixel-based ratios of circumscribed area to the normalized residue proportion were calculated for the valleculae (NRRSv) and pyriform sinuses (NRRSp) when the hyoid bone returned to its lowest position after a swallow. Raters for NRRS analysis were speech language pathologists who did not participate in data collection and were blinded to pre- or post-operative and neck dissection status. Inter-rater reliability correlations were calculated for NRRS ratings of both vallecular and pyriform sinus residue.

Table 2.

Randomized group characteristics. CND = centralized neck dissection; TT = total thyroidectomy without neck dissection

Group Characteristics	Number (%) patients n = 32
Treatment Groups
CND	15 (47%)
TT	17 (53%)
Sex
CND
Female	11 (73%)
Male	4 (27%)
TT
Female	12 (71%)
Male	5 (29%)
Age	Mean age in years (SD)
CND	50 (14)
TT	48 (14)

Within the CND and TT treatment groups, mean score and standard deviation (SD) were calculated for EAT-10 as well as NRRSv, NRRSp, and PAS with each consistency. Two-factor, mixed-design ANOVA tests were performed to assess statistically significant differences in mean scores for EAT-10, PAS for each consistency, and NRRS for each consistency between the TT and CND groups as well as analysis for effect of time. Given that no significant differences were found between treatment groups, the treatment groups were combined for post-hoc mean comparison between pre- and post-operative time points using Wilcoxon signed-rank test. Spearman correlation coefficients were calculated among EAT-10, PAS, and NRRS for the combined post-operative treatment group. Inter-rater reliability for pre- and post-operative NRRS measures was calculated using Spearman correlation coefficients. The α-level for determining statistical significance was < 0.05.

Results

Data collected from 52 patients enrolled in Clinical Trial NCT02138214 were considered for inclusion in this study (Fig. 1). Seven patients withdrew from the trial prior to randomization, 10 subjects were excluded based on intraoperative evaluation, and 3 were lost to follow-up. Of the 32 remaining patients, 17 were included in the TT group and 15 in the CND group. The TT group contained 12 females with an average age of 50.25 (± 13.21) years and 5 males with an average age of 41.80 (± 14.13) years. The CND group contained 11 females with an average age of 47.45 (± 13.70) years and 4 males with an average age of 56.25 (± 12.71) years.

Fig. 1.

Consort diagram

EAT-10

Mean EAT-10 score of the CND group was 1.80 (± 3.40), compared to the mean score of the TT group was 3.38 (± 5.30). Analysis of mean scores (Table 3) yielded no statistically significant difference between the CND and TT groups (p = 0.2406) with no significant effect of time between pre- and post-operative study points (p = 0.1090). In a post-hoc comparison of pre- and post-operative time points for the combined treatment groups (Table 4), no significant difference was found (p = 0.1374).

Table 3.

Comparison of mean swallow differences by treatment group and interactions with time points. p < 0.05; [1] no ANOVA test due to all values = 1

Factor	Group	Mean (SD)	P-value
EAT-10	CND	1.80 (3.4)	0.2406
EAT-10 x time	TT	3.38 (5.3)	0.1090
PAS Thin	CND	2 (1)	0.4465
PAS Thin x time	TT	2 (1)	0.8127
PAS Pudding	CND	1 (0)	none1
PAS Pudding x time	TT	1 (0)	none1
PAS Solid	CND	1 (0)	none1
PAS Solid x time	TT	1 (0)	none1
NRRSv Thin	CND	0.02 (0.04)	0.2229
NRRSv Thin x time	TT	0.01 (0.02)	0.2094
NRRSp Thin	CND	0.01 (0.03)	0.0908
NRRSp Thin x time	TT	0.0002 (0.0006)	0.7508
NRRSv Pudding	CND	0.03 (0.06)	0.9303
NRRSv Pudding x time	TT	0.03 (0.09)	0.5876
NRRSp Pudding	CND	0.002 (0.01)	0.8783
NRRSp Pudding x time	TT	0.002 (0.005)	0.6726
NRRSv Solid	CND	0.07 (0.11)	0.0520
NRRSv Solid x time	TT	0.01 (0.02)	0.2767
NRRSp Solid	CND	0.01 (0.05)	0.6432
NRRSp Solid x time	TT	0.01 (0.02)	0.5399

CND = centralized neck dissection; TT = total thyroidectomy without neck dissection; EAT-10 = Eating Assessment Tool; PAS = Penetration Aspiration Scale; NRRSv = Normalized Residue Ratio Scale – valleculae; NRRSp = Normalized Residue Ratio Scale – pyriform sinus

Table 4.

Comparison of mean swallow differences for combined groups at post-operative time point. p < 0.05; [1] no ANOVA test due to all values = 1

Factor	Pre-operative (mean ± SD)	Post-operative (mean ± SD)	P-value
EAT-10	1.91 ± 3.22	3.28 ± 5.31	0.1374
PAS
Thin	1.72 ± 1.05	1.66 ± 1.41	0.7588
Pudding	1.00 ± 0.00	1.00 ± 0.00	none1
Solid	1.00 ± 0.00	1.00 ± 0.00	none1
NRRS
NRRSv Thin	0.02 ± 0.04	0.01 ± 0.01	0.8650
NRRSp Thin	0.01 ± 0.03	0.00 ± 0.01	0.5830
NRRSv Pudding	0.03 ± 0.06	0.03 ± 0.09	0.8599
NRRSp Pudding	0.00 ± 0.00	0.00 ± 0.01	0.9097
NRRSv Solid	0.03 ± 0.07	0.06 ± 0.11	0.2455
NRRSp Solid	0.01 ± 0.02	0.01 ± 0.06	0.3750

EAT-10 = Eating Assessment Tool; PAS = Penetration Aspiration Scale; NRRSv = Normalized Residue Ratio Scale – valleculae; NRRSp = Normalized Residue Ratio Scale – pyriform sinus

PAS

For thin liquid PAS, mean score of the CND group was 1.83 (± 1.44), and the TT group was 1.56 (± 1.02). ANOVA comparison of mean scores yielded no statistically significant difference for treatment groups (p = 0.4465) or for time effect (P = 0.8127). For the PAS with pudding and solid consistencies, all scores equaled 1.00 for the CND and TT groups as well as the pre-operative and post-operative time points. Due to this uniformity, ANOVA comparisons were not performed. In analysis of the combined treatment groups for pre- and post-operative comparison for thin liquids, no significant differences were found (p = 0.7588).

NRRS

Inter-rater correlations for NRRS measures between two raters are shown in Table 5. Correlations between raters for vallecular stasis measures were very strong, with r > 0.82046 across three consistencies in both pre- and post-operative conditions. Correlations for pyriform sinus stasis were slightly reduced in strength, with 0.61394 < r > 0.74741 representing a strong correlation for all measures except pre-operative pudding stasis that showed a weak relationship of r = -0.16563.

Table 5.

Inter-rater correlations for NRRS measures. p < 0.05

NRRS measure	Spearman Correlation Coefficient	P-value
Valleculae
Pre-op Thin	1.00000	< .0001
Post-op Thin	0.82046	0.0036
Pre-op Pudding	0.90968	0.0003
Post-op Pudding	0.83226	0.0028
Pre-op Solid	0.86702	0.0012
Post-op Solid	1.00000	< .0001
Pyriform Sinus
Pre-op Thin	0.61394	0.0590
Post-op Thin	0.66021	0.0377
Pre-op Pudding	-0.16563	0.6475
Post-op Pudding	0.74536	0.0133
Pre-op Solid	0.74536	0.0133
Post-op Solid	0.74741	0.0130

Thin liquid

Mean NRRSv score with thin liquid consistency in the CND group was 0.02 (± 0.04) and 0.01 (± 0.02) for the TT group. ANOVA comparison of mean scores yielded no statistically significant difference between the treatment groups (p = 0.2229) or for time effect within treatment group (p = 0.2094). For the NRRSp with thin liquid consistency, mean score of the CND group was 0.01 (± 0.03), and mean score of the TT group was 0.0002 (± 0.0006). ANOVA comparison of mean scores yielded no statistically significant difference between treatment groups (p = 0.0908) or time effect (P = 0.7508). Pre- and post-operative comparison for the combined groups revealed no significant differences between time points for NRRSv and NRRSp (p = 0.8650 and 0.5830, respectively).

Pudding

Mean NRRSv scores for pudding consistency in the CND group was 0.03 (± 0.06), and the TT group was 0.03 (± 0.09). ANOVA comparison of mean scores yielded no statistically significant difference between the treatment groups (P = 0.9303) or time effect (P = 0.5876). For the NRRSp with pudding consistency, mean score of the CND group was 0.002 (± 0.01), and mean score of the TT group was 0.002 (± 0.005). Mean score of the pre-operative time point was 0.002 (± 0.005), and mean score of the post-operative time point was 0.002 (± 0.01). ANOVA comparison of mean scores yielded no statistically significant difference between the CND and TT groups (P = 0.8783) or between the pre-operative and post-operative time points (P = 0.6726).

Solid

For the NRRSv with solid consistency, mean score for the CND group was 0.07 (± 0.11), and mean score for the TT group was 0.01 (± 0.02). Mean score for the pre-operative time point was 0.03 (± 0.07), and mean score for the post-operative time point was 0.06 (± 0.11). ANOVA comparison of mean scores yielded no statistically significant difference between the CND and TT groups (P = 0.052) or between the pre-operative and post-operative time points (P = 0.2767).

For the NRRSp with solid consistency, mean score for the CND group was 0.01 (± 0.05), and mean score for the TT group was 0.01 (± 0.02). Mean score for the pre-operative time point was 0.01 (± 0.02), and mean score for the post-operative time point was 0.01 (± 0.06). ANOVA comparison of mean scores yielded no statistically significant difference between the CND and TT groups (P = 0.6432) or between the pre-operative and post-operative time points (P = 0.5399).

Discussion

The hypotheses of this study postulated that patients who underwent total thyroidectomy would show worse measures of swallowing efficiency, penetration/aspiration ratings, and patient-rated swallowing outcomes following surgery, with prophylactic neck dissection diminishing swallow outcomes. Our findings did not support our hypotheses. Among the analyzed variables in this investigation, no clinically relevant penetration/aspiration or pharyngeal residue volumes were identified to account for the 80% of clinical trial participants who reported symptoms of dysphagia 2 weeks after thyroidectomy in qualitative analysis [12]. These findings are consistent with a previous systematic review reflecting disparity between patient-perceived dysphagia and instrumental assessment data [3]. A possible explanation for this disparity could be that categorical themes that emerged in the previous qualitative analysis [12], including pain, lump sensation, and swallowing effort, are sensory in nature and not captured on imaging studies. The “lump in throat” complaint was the predominant complaint reported at both 6 weeks (27%, n = 7) and 6 months (8%, n = 2).¹² Furthermore, randomization of central neck dissection in this trial with no significant differences in outcomes between the two groups validates a previous systematic review suggesting no additional risk of dysphagia associated with extended resection [11].

Patient-rated EAT-10 scale scores collected at pre- and post-operative time points in patients undergoing total thyroidectomy did not capture significant change in patient-reported function. Furthermore, this analysis revealed no significant differences in EAT-10 scores between those participants randomized to undergo CND and participants who did not. The mean scores for both surgical groups across operative time points fell at or below the normative threshold score of 3 or higher associated with clinically relevant dysphagia [15]. These results are similar to EAT-10 scale scores collected among patients undergoing total thyroidectomy for symptomatic goiter, where mean scores before, 1 month following and 3 months following surgery fell below the normative threshold [17].

PAS ratings representing depth of airway invasion and sensory response during swallowing showed no significant differences between time points or randomized neck dissection status. PAS ratings of 4 or higher would hold clinical relevance to respiratory health, representing invasion to the vocal folds or below [14]. The mean PAS values of falling below 2 reported in this study suggest that most participants experienced no invasion or only shallow penetration into the airway, unlikely to be associated with respiratory health consequences.

Measurements of pharyngeal residue in the valleculae and pyriform sinuses revealed no significant differences between randomized neck dissection groups or time points. Published validation of NRRS measures have shown that NRRSp values greater than 0.09 are associated with 2.07 times relative risk of penetration or aspiration [18]. Measures in these two group comparisons fell well below this clinically relevant threshold.

There are limitations that must be considered in the interpretation of the study. The small trial sample who had imaging studies available for analysis may have impacted the statistical power of the study results. In addition, the EAT-10 tool does not capture patient ratings of sensory outcomes other than pain, confining sensory data related to quality of life that was available for analysis. Future investigation of post-thyroidectomy dysphagia should focus on measurement of perceived swallowing discomfort and effort that may account for the disparity between functional status and post-operative swallowing complaints. Videofluoroscopic swallowing imaging may be limited in capturing compensatory effort or fatigue in swallowing function. Improved understanding of the patient’s swallowing experience would better inform pre-operative education and post-operative care that guides patients through thyroidectomy surgeries.

The implications of these findings suggest that patient perceptions of dysphagia following thyroidectomy may be sensory in nature but dissociated from motor aspects of swallowing. The absence of post-operative aspiration or pharyngeal residual offers reassurance that swallowing safety and efficiency are not only adequate prior to surgery, but also recovered by two weeks following surgery to the baseline performance. Patients distressed by perceived dysphagia as they face surgical resection of papillary thyroid cancer can be offered comfort in the likelihood that sensory symptoms are unlikely to threaten respiratory health or swallowing capability.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by all authors. The first draft of the manuscript was written by M.A.K. and D.R.The final draft of the manuscript was written by M.A.K.All tables and figures were prepared by M.A.K. All authors read and approved the final manuscript.

Funding

This work was funded in part by NIH R01CA176911 (Sippel, Connor, PIs) and the Diane M. Bless Endowed Chair.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained for each participant under the approval of the University of Wisconsin – Madison Health Sciences Institutional Review Board (IRB 2014 − 0391).

Consent to participate

Informed consent was collected for all participants enrolled in the IRB approved studies listed above.