Muscle activity and kinematics show different responses to recurrent laryngeal nerve lesion in mammal swallowing

Abstract

Understanding the interactions between neural and musculoskeletal systems is key to identifying mechanisms of functional failure. Mammalian swallowing is a complex, poorly understood motor process. Lesion of the recurrent laryngeal nerve, a sensory and motor nerve of the upper airway, results in airway protection failure (liquid entry into the airway) during swallowing through an unknown mechanism. We examined how muscle and kinematic changes after recurrent laryngeal nerve lesion relate to airway protection in eight infant pigs. We tested two hypotheses: 1) kinematics and muscle function will both change in response to lesion in swallows with and without airway protection failure, and 2) differences in both kinematics and muscle function will predict whether airway protection failure occurs in lesion and intact pigs. We recorded swallowing with high-speed videofluoroscopy and simultaneous electromyography of oropharyngeal muscles pre- and postrecurrent laryngeal nerve lesion. Lesion changed the relationship between airway protection and timing of tongue and hyoid movements. Changes in onset and duration of hyolaryngeal muscles postlesion were less associated with airway protection outcomes. The tongue and hyoid kinematics all predicted airway protection outcomes differently pre- and postlesion. Onset and duration of activity in only one infrahyoid and one suprahyoid muscle showed a change in predictive relationship pre- and postlesion. Kinematics of the tongue and hyoid more directly reflect changes in airway protections pre- and postlesion than muscle activation patterns. Identifying mechanisms of airway protection failure requires specific functional hypotheses that link neural motor outputs to muscle activation to specific movements.

NEW & NOTEWORTHY Kinematic and muscle activity patterns of oropharyngeal structures used in swallowing show different patterns of response to lesion of the recurrent laryngeal nerve. Understanding how muscles act on structures to produce behavior is necessary to understand neural control.

INTRODUCTION

Animal movements are the observable output of the sequentially connected activities of the nervous system and musculoskeletal system and associated processes (motor pattern production, muscle activation, and kinematics) (Biewener and Daniel 2010; German et al. 2017). Nervous system activity and musculoskeletal function can be analyzed separately for their impact on behavior, particularly how behavior changes when the system is disturbed. Examining each of these levels in isolation, however, limits our ability to measure how the levels interact. This leads to treating the rest of the system as a “black box,” which assumes that the effect of the perturbation will propagate in a simple, easily predicted manner from the system of interest through the other levels of the system to the end point behavior. Yet, each system has distinctive physiological properties, for example, the refractory period of muscles (Hylander and Johnson 1993), or the feedforward mechanisms of the neural systems (Criscimagna-Hemminger et al. 2010; Humbert et al. 2013), that affect how they can interact with each other. In addition, behavior produces feedback through mechanical and neurological pathways that may itself affect the physiological response (German et al. 2017; Holman et al. 2013b). The links and interactions that exist between nervous system activity and musculoskeletal function, and how they influence behavioral outcomes, have the potential to enhance our understanding of how behavior is controlled.

Mammalian swallowing is a complex behavior involving over 20 paired muscles innervated by at least seven cranial and cervical spinal nerves (German et al. 2017; Miller 1986; Thexton and Crompton 1998). These muscles can contract eccentrically, concentrically, or isometrically within and between swallows (Holman et al. 2012; Konow et al. 2010). The anatomical structures (tongue, pharynx, soft palate) that produce swallowing are deformable soft tissues with changing shapes and mechanical properties whose physical actions are difficult to measure (Gould et al. 2017; Orsbon et al. 2018; Thexton et al. 2004). The pathways and connections between brainstem (Jean 2001; Thexton et al. 2009) and cortical centers (Hamdy et al. 1996; Humbert and German 2013; Michou and Hamdy 2009) involved in swallowing are incompletely understood. In particular, the role of sensation from different parts of the oropharynx and larynx and their impact on performance remain unknown (Barlow et al. 2014; Gould et al. 2016; Holman et al. 2013b; Steele and Miller 2010).

Our incomplete understanding of both the biomechanics and the neural control of swallowing limits our ability to determine what drives performance and performance failure in swallowing. Airway protection—the ability to pass the bolus over the opening of the trachea into the esophagus safely and quickly—is a key feature of mammalian swallowing with clinical (Steele and Cichero 2014) and evolutionary relevance (Smith 1992; Zhou et al. 2019). However, what behavioral characteristics of swallowing are crucial to achieving airway protection remains unclear. The recurrent laryngeal nerve (RLN) is a mixed sensory and motor nerve that innervates the mucosal lining of the lower larynx and the muscles that move the vocal cords, structures outside the path of the bolus in a normal swallow. Despite its targets not being anatomically implicated in swallowing, unilateral lesion of the recurrent laryngeal nerve (RLN) leads to increased aspiration—entry of food or liquid into the airway (Gould et al. 2015; Périé et al. 1998). Indeed, tongue kinematics (Gould et al. 2016), timing of laryngeal vestibule closure (Gross et al. 2018), bolus size and shape (Gould et al. 2017), and timing and duration of the activity of hyoid and tongue muscles (DeLozier et al. 2018) are all altered after RLN lesion. Because many of these structures are not innervated by the RLN or moved by muscles innervated by the RLN, these results point to a pervasive effect of RLN lesion on central sensory motor integration of swallowing behavior. However, despite all these documented effects of RLN lesion on swallowing function, the mechanism of airway protection failure in RLN lesion remains elusive (Newman et al. 2001).

In this study, we use our validated infant pig model to integrate the interactions and relationships among 1) RLN lesion, 2) muscle activity, 3) kinematics of oropharyngeal structures, and 4) airway protection as a behavioral outcome. Our main goal is to determine the links from physiology to performance following the lesion of RLN.

We address this problem with two complementary approaches. First, we examine how electromyography (EMG) and kinematics differ with failure or success of airway protection in lesioned and intact animals. This approach tests for differences in kinematics and muscle activation at different levels of performance (airway protection) and lesion status. Our second approach asks whether muscle activity and kinematics can predict the performance outcome in lesion and intact individuals, as measured by airway protection outcomes.

We delineate our two approaches with two hypotheses, as follows:

• Hypothesis 1: Both EMG and kinematic variables will differ between safe and unsafe swallows in both lesion and intact animals, reflecting the connection between neurologically controlled motor function and kinematics.

• Hypothesis 2: Kinematics and muscle function will predict airway protection outcomes. Lesion will modify the predictive relationship in both sets of variables.

MATERIALS AND METHODS

Animals

Eight infant pigs (Yorkshire/Landrace) were purchased at between 3 and 11 days of age (Shoups Farm, Wooster, OH). Pigs were acclimated over 36 h to drink standard pig milk replacer (Solustart Pig Milk Replacement, Land o’ Lakes, Arden Mills, MN) from a bottle with a modified pig nipple (NASCO Farm and Ranch, Fort Atkinson, WI). Because of variability in electrode quality and marker placement, we had four pigs with both EMG and kinematic data, two pigs with just EMG data, and two with just kinematic data.

All animal work was approved by the Northeast Ohio Medical University Institutional Animal Care and Use Committee (NEOMED IACUC) and followed Association for the Assessment and Accreditation of Laboratory Animal Care and United States Department of Agriculture Guidelines for Care and Use of Animals in Research.

Surgery

All procedures were performed in accordance with approved NEOMED IACUC protocols 13-011 and 16-007. Once animals were drinking milk effectively from the bottle, they underwent two procedures. In a nonsterile intraoral procedure, while in a deep plane of isoflurane anesthesia, we implanted radio-opaque markers along the midline into the tongue, soft palate, and gingiva of the hard palate, using a 19-gauge needle. A radio-opaque microhematological clip (Weck Ligation Solutions, NC) was attached to the epiglottis to track its movements. In a subsequent fully sterile surgery, under anesthesia (2%–5% intubated isoflurane), we implanted bipolar, fine-wire electrodes directly into hyoid musculature. The following muscles were identified visually based on anatomical dissection before bilateral implantation of electrodes: geniohyoid, genioglossus, mylohyoid, stylohyoid, cricothyroid, and thyrohyoid. None of these muscles is innervated by the RLN, so no postlesion loss of function is expected. During this surgery, radio-opaque markers were also sutured to the hyoid and thyroid. Analgesia [buprenorphine (0.1 mg/kg) and loxicom (0.4 mg/kg)] was provided before and for 48 h after surgery. Animals were monitored for any sign of discomfort after surgery every 3 to 5 h.

After surgery and marker implantation, animals were recorded in lateral view drinking in front of the fluoroscope every 4–8 h for 48 h before undergoing a second surgery, again under intubated isoflurane anesthesia. In this surgery, the right recurrent laryngeal nerve was identified in the neck, running along the trachea, just caudal to its point of entry into the larynx. Taking care not to displace any electrodes, we tied the nerve with suture in two places, crushed it with microhematological clips in two places, and then removed a 1–2-mm section of the nerve between the two sutures. The cut nerve ends were displaced to prevent re-innervation. Animals were recorded for a further 72–96 h postlesion. At the end of the experiment, while in a deep plane of isoflurane anesthesia, animals were euthanized by intracardiac injection of pentobarbital. An investigator who had not performed the surgeries performed necropsy to confirm electrode position and nerve lesion. All procedures were identical to those described in our previous studies (DeLozier et al. 2018; Gould et al. 2016).

Videofluoroscopy and Electromyography

Animals were recorded in lateral view drinking pig milk replacement formula mixed with barium (E-Z Paque Barium Sulfate, E-Z-EM Inc., NY) in front of a fluoroscope (85 kVp, 4.0 MA, GE9400 C-arm) attached to a high-speed (100 fps) digital video camera (Xcitex 1M, Xcitex, Woburn, MA). Simultaneously, preamplified EMGs were captured at 10 kHz using an MA-300 EMG system (Motion Laboratory Systems) with a 20-Hz to 2-kHz bandpass filter and a 60-Hz notch filter. EMG signals were recorded on a digital chart recorder (Powerlab 30/16, ADInstruments). EMG and videofluoroscopic recordings were synchronized using a manually triggered square-wave pulse-trace recording. Animals were allowed to feed unrestrained (other than EMG tethering) until satiation. We recorded for entire feeding bouts (i.e., from when an animal latched onto the nipple to when it voluntarily released it).

Animals were recorded three to four times between EMG and RLN lesion surgeries to obtain prelesion data. After lesion surgery, all animals were recorded at regular time intervals for 92 h postsurgery.

Measurement of Swallow Performance

All swallows from entire feeding sequences (between 20 and 100 swallows depending on age and individual) were scored for airway protection performance. An investigator, blinded to lesion status, scored swallows using the infant mammalian penetration-aspiration score (IMPAS; Holman et al. 2013a). The IMPAS score is a validated, ordinal scale of airway protection, which rates swallows on a seven-point scale from 1 (complete airway protection) to 7 (silent aspiration) (Table 1). Prior to scoring swallows, investigators trained on the same set of scored swallows until intra- and interrater reliability scores of 90% were reached.

Table 1.

Description of the IMPAS scale, after Holman et al. (2013a)

Score	What happens
1	Normal swallow
2	Some penetration that is cleared during the swallow
3	Some penetration that is not cleared during the swallow
4	A lot of penetration that is not cleared during the swallow
5	Aspiration with a successful attempt to clear
6	Aspiration with an unsuccessful attempt to clear
7	Aspiration with no attempt to clear

IMPAS, infant mammalian penetration-aspiration score.

Quantification of EMG

EMG data were visually assessed for quality of signals, looking for signs of damaged, broken, or misplaced electrodes. Sequences of 10–30 contiguous swallows, identified from thyrohyoid burst (Thexton et al. 2007), were extracted and then passed through a custom program (Sequencer; Thexton 1996), in which baseline corrects, rectifies, and integrates the signal. A custom MATLAB (MathWorks) script was then used to calculate duration of muscle activity and timing of peak muscle activity relative to peak left thyrohyoid firing for that swallow (DeLozier et al. 2018). Because of variability in electrode signal and durability, the same number of individuals could not be included for each muscle. Sample sizes by electrode and individual pre- and postlesion are provided in Supplemental material (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.12933383.v1).

Using the square-wave synchronization pulse, sequences of swallows from the video sequences and sequences of swallows from the EMG data (as indicated by time of thyrohyoid activity) were set to the same time base. Thus, IMPAS scores were assigned to the EMG swallow variables.

Quantification of Kinematics

Sequences of ∼20 swallows were extracted from feeding bouts for kinematic analysis. Swallows were selected from the middle of the feeding bout when swallow rate is most regular (Gierbolini-Norat et al. 2014). Radio-opaque markers were tracked in two dimensions using marker-tracking software (ProAnalyst, Xcitex, Woburn, MA). The resulting x, y coordinates were all aligned and normalized to hard palate length by scaling, rotating, and translating each frame in series of swallows to the two hard palate markers, which were used to define a horizontal axis. The transformed x, y coordinates for the tongue, epiglottis, and hyoid track movements of those structures were relative to the hard palate, eliminating the effects of head movements.

From the digitized traces of marker movements, swallow duration was defined as the time from beginning of epiglottal inversion to the time of return of the epiglottis to its resting position (Ding et al. 2013; Gould et al. 2016). Within each swallow duration, we calculated three metrics: posterior tongue ratio, middle tongue ratio, and time of maximum hyoid elevation. The posterior tongue ratio is the ratio of the distance traveled by the posterior tongue marker in the time it takes for the epiglottis to reach its caudal most point to the total distance traveled by the posterior tongue marker during a swallow (Gould et al. 2016). The middle tongue ratio is the same as the posterior tongue ratio but calculated for the middle tongue marker. The time of maximum hyoid elevation is the time during the swallow when the hyoid marker has moved furthest from its initial position. These metrics were selected based on the previous work, indicating that timing of movements is most affected by recurrent laryngeal nerve lesion (Gould et al. 2016; Gross et al. 2018).

Testing the Impact of Lesion on Airway, Kinematic, and EMG Variables (Hypothesis 1)

To test for the effect of lesion and airway protection outcomes on kinematics and EMG variables (hypothesis 1), we used a mixed-model ANOVA approach with individual animal as a random factor to control for interindividual variation. Because of unequal numbers of electrodes, left and right muscles were tested separately. The fixed factors were IMPAS score, lesion status, and their interaction. Because IMPAS scores 3 through 6 were rare or absent, only IMPAS scores of 1, 2, and 7 were retained for analysis. This data selection represented the most comprehensive, thus conservative test, and we could run with our data. We ran the model separately for each of the kinematic (posterior tongue ratio, middle tongue ratio, time of maximum hyoid elevation) and EMG (time and duration of muscle activation) response variables. Where significant interactions were recovered, we used a priori planned contrast tests to test for pairwise differences. To limit the number of postcontrast tests, we only test for differences between IMPAS 1 and IMPAS 7. We tested for the following: differences between IMPAS 1 and IMPAS 7 within lesion and control, differences in IMPAS 1 between lesion and control, differences in IMPAS 7 between lesion and control, and changes in the difference between IMPAS 1 and IMPAS 7 from lesion to control (difference of differences). These analyses were performed in R (R Core Team 2015) using the packages lmer and emmeans.

Predictive Value of Kinematic and EMG Variables for IMPAS as a Measure of Performance (Hypothesis 2)

To test the power of kinematic and EMG variables to predict performance outcome, measured by IMPAS, for both lesion and control individuals, we used logistic regression (Grund and Sabin 2010; Hoo et al. 2017; Metz 1978). The small number of scores of 2, 3, and 4 limited the power of an analysis using all IMPAS scores. There were no scores of 5 or 6, which are rare to nonexistent in this model (Gould et al. 2015, 2016). Thus, we used only scores of 1 (normal and safe swallow) and 7 (unrecovered aspiration) as predicted outcomes. The model included a physiologic variable (kinematic, timing EMG, or duration EMG) and treatment (control or lesion) and their interaction:

$$IMPAS=constant+\beta 1^{\text{*\hspace{0.17em}}}\left(physiologicvariable\right)+\beta 2^{\text{*\hspace{0.17em}}}\left(treatment\right)+\beta 3^{\text{*\hspace{0.17em}}}\left(physiologicva{r}^{\text{*\hspace{0.17em}}}treatment\right)$$

The parameters of the regression (β1, β2, and β3) indicate whether a specific variable or variable interaction contributes to the ability of the model to distinguish between outcomes, in this case IMPAS = 1 or IMPAS = 7. We used IMPAS = 1 as the baseline in these analyses. Thus, we tested whether the value of a physiologic variable (either kinematic or EMG, e.g., the maximum movement of the hyoid bone during a swallow) predicts the IMPAS outcome. The interaction term tests whether the relationship between that continuous physiologic variable and treatment state (intact/lesion) predicts the IMPAS outcome.

We report odds ratio (OR) estimates for each parameter and the area under the receiver operating characteristics (ROC) curve. The OR measures the relationship between two variables. Thus, an OR = 1 means that two variables are independent, and the outcome of the predicted variable (here, IMPAS) does not depend on the state of the predictor variable (EMG, kinematics, or lesion status). An OR > 1 means that the occurrence or higher value of the predictor raises the odds that the outcome will occur. An OR < 1 means that the occurrence/lower value of the predictor reduces the odds that the outcome will occur. The ROC curve is a map of how changes in the false-positive rate relate to changes in the true-positive rate. Thus, high area under the curve suggests that the test or predictive ability is high in specificity and sensitivity.

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

RESULTS

Hypothesis 1a: The Effects of Lesion and IMPAS Are Different in Kinematic Variables and in EMG Variables

All kinematic variables vary with lesion and airway protection status.

For all three kinematic response variables, the interaction of IMPAS score with lesion status was significant in the type III mixed-model ANOVA (Table 2 and Fig. 1). Post hoc planned contrast tests indicated that in control recordings only, values of posterior tongue ratio, middle tongue ratio, and time of maximum hyoid elevation are higher for swallows with silent aspiration (IMPAS 7) than for safe swallows (IMPAS 1) (Table 3). The difference in posterior tongue ratio, middle tongue ratio, and time of maximum hyoid elevation between swallows with silent aspiration and safe swallows is greater prelesion than postlesion. Posterior tongue ratio and middle tongue decrease in swallows with silent aspiration postlesion. Posterior tongue ratio and middle tongue in safe swallows increase postlesion.

Table 2.

Results of mixed-model ANOVA for middle tongue ratio, posterior tongue ratio, and time of maximum hyoid elevation

Kinematic Variable	Factor	F (numerator df, denominator df)	P Value
MT ratio	Treatment	24.78 (1, 412.01)	<0.001
	IMPAS score	10.67 (2, 412.73)	<0.001
	Treatment:IMPAS	5.76 (2, 410.25)	<0.01
PT ratio	Treatment	8.75 (1, 326.76)	<0.01
	IMPAS score	10.11 (2, 327.33)	<0.001
	Treatment:IMPAS	14.07 (2, 325.7)	<0.001
Time max hyoid elevation	Treatment	0.5 (1, 415.43)	0.48
	IMPAS score	5.27 (2, 412.97)	<0.01
	Treatment:IMPAS	10.19 (2, 413.08)	<0.001

Bold F statistics and P values indicate significant factors. df, Degrees of freedom; IMPAS, infant mammalian penetration-aspiration score; max, maximal; MT, middle tongue; PT, posterior tongue.

Fig. 1.

Least squares means with 95% confidence intervals of middle tongue (MT) ratio (top), posterior tongue (PT) ratio (middle), and time of hyoid elevation (bottom) for IMPAS scores 1, 2, and 7 pre- and postlesion. Brackets indicate significant pairwise contrast tests. IMPAS, infant mammalian penetration-aspiration score.

Table 3.

Result of contrast tests

Contrast	Variable	Control	Lesion
Difference between IMPAS 1 and IMPAS 7	PT ratio	<0.0001	ns
	MT ratio	<0.0001	ns
	Hyoid elevation	0.0004	ns
Difference between C and L for IMPAS 1	PT ratio		0.009
	MT ratio		ns
	Hyoid elevation		0.001
Difference between C and L for IMPAS 7	PT ratio		0.001
	MT ratio		0.0003
	Hyoid elevation		ns
Difference C and L for diff ofIMPAS 1–7	PT ratio		0.0001
	MT ratio		0.01
	Hyoid elevation		0.005

Bold control and lesion P values indicate significant factors. C, control; diff, difference; IMPAS, infant mammalian penetration-aspiration score; L, lesion; MT, middle tongue; ns, not significant; PT, posterior tongue.

Limited differences in duration of muscle activation exist between swallows with different lesion and airway protection statuses.

Figure 2 shows representative traces of the raw EMG data. The relationship between lesion and airway protection and duration of hyoid muscle activity was weak (Table 4). The thyrohyoid muscle (left and right) fired significantly longer postlesion (Fig. 3, left column). For the left thyrohyoid, duration of activation was longer in swallows with silent aspiration (Fig. 3, top right). A significant lesion status and IMPAS interaction were found for duration of activation of the right stylohyoid. Post hoc contrast tests indicated a significant increase in duration of activation of muscle in safe swallows only postlesion in the right stylohyoid (Fig. 3, bottom right).

Fig. 2.

Representative EMG traces prelesion (left) and postlesion (right) from two individuals covering all muscles sampled in this analysis. Dotted lines indicate onset of swallows as determined by onset of the right thyrohyoid muscle.

Table 4.

Results of main effects and interaction ANOVA on type III linear mixed models on duration of activation of hyoid muscles during swallowing

Muscle	Factor	F (numerator df, denominator df)	P Value
Left thyrohyoid	Treatment	52.81 (1, 348)	<0.001
(LTH)	IMPAS score	5.32 (2, 348)	0.005
	Treatment:IMPAS	2.19 (2, 348)	0.113
Right thyrohyoid	Treatment	59.06 (1, 540)	<0.001
(RTH)	IMPAS score	1.43 (2, 540)	0.24
	Treatment:IMPAS	2.08 (2, 540)	0.125
Left genioglossus	Treatment	0.96 (1, 227)	0.329
(LGG)	IMPAS score	1.52 (2, 227)	0.221
	Treatment:IMPAS	1.8 (2, 227)	0.167
Right genioglossus	Treatment	<0.01 (1, 244)	0.971
(RGG)	IMPAS score	0.3 (1, 244)	0.741
	Treatment:IMPAS	0.15 (2, 244)	0.863
Left geniohyoid	Treatment	0.36 (1, 85)	0.548
(LGH)	IMPAS score	1.39 (2, 85)	0.256
	Treatment:IMPAS	1.88 (2, 85)	0.16
Right stylohyoid	Treatment	6.5 (1, 176)	0.012
(RStyH)	IMPAS score	0.47 (2, 176)	0.628
	Treatment:IMPAS	3.2 (2, 176)	0.043
Left cricothyroid	Treatment	0.58 (1, 207)	0.445
(LCT)	IMPAS score	1.41 (2, 207)	0.247
	Treatment: IMPAS	0.45 (2, 207)	0.633

Bold F statistics and P values indicate significant differences. IMPAS, infant mammalian penetration aspiration score.

Fig. 3.

Least squares means with 95% confidence intervals of significant main effects or interactions (Table 5) for ANOVA on type III linear mixed models of lesion status, IMPAS score, and their interaction on duration of muscle activity. Bracket indicates significant pairwise differences for interactions. IMPAS, infant mammalian penetration-aspiration score; LTH, left thyrohyoid; RTH, right thyrohyoid; RStyH, right stylohyoid.

Timing of muscle activation differs among swallows with different airway protection and lesion statuses.

The time of activation of the left thyrohyoid and right and left genioglossus relative to the activation of the right thyrohyoid was significantly earlier postlesion (Table 5 and Fig. 4). The time of activation of the left thyrohyoid, right genioglossus, left stylohyoid, and right cricothyroid was significantly earlier in swallows with aspiration than safe swallows (Table 5 and Fig. 4). There was a significant interaction of lesion status and IMPAS score for the left cricothyroid. Post hoc contrast test showed that the time of onset of activation of left cricothyroid was significantly earlier postlesion in safe swallows only (Table 5 and Fig. 4).

Table 5.

ANOVA table for main effect and interaction on type III linear mixed model of the effect of lesion status and IMPAS score on timing of activation of hyoid musculature in swallowing

Muscle	Factor	F (numerator df, denominator df)	P Value
Left thyrohyoid	Treatment	13.9 (1, 311)	<0.001
(LTH)	IMPAS score	4.78 (2, 311)	0.009
	Treatment:IMPAS	1.1 (2, 311)	0.334
Left genioglossus	Treatment	27.266 (1, 264)	<0.001
(LGG)	IMPAS score	2.94 (2, 264)	0.055
	Treatment:IMPAS	2.77 (2, 264)	0.064
Right genioglossus	Treatment	7.53 (1, 274)	<0.001
(RGG)	IMPAS score	6.08 (2, 274)	0.003
	Treatment:IMPAS	1.72 (2, 274)	0.181
Left geniohyoid	Treatment	0.54 (1, 162)	0.463
(LGH)	IMPAS score	2.13 (2, 162)	0.122
	Treatment:IMPAS	1.37 (1, 162)	0.258
Left stylohyoid	Treatment	0.29 (1, 144)	0.591
(LStyH)	IMPAS score	3.73 (2, 144)	0.026
	Treatment:IMPAS	1.33 (2, 144)	0.267
Right cricothyroid	Treatment	0.45 (1, 154)	0.505
(RCT)	IMPAS score	3.47 (2, 154)	0.034
	Treatment:IMPAS	1.97 (2, 154)	0.143
Left cricothyroid	Treatment	4.32 (1, 161)	0.039
(LCT)	IMPAS score	0.87 (2, 161)	0.419
	Treatment:IMPAS	4.97 (2, 161)	0.008

Individual was a random factor. Bold F statistics and P values indicate significant effects. IMPAS, infant mammalian penetration aspiration score.

Fig. 4.

Least squares means with 95% confidence intervals for significant main effects and interactions for timing of muscle activation relative to right thyrohyoid for ANOVA on type III linear mixed models of lesion status, IMPAS score, and their interaction. Bracket indicates significant pairwise differences as determined by post hoc tests on interaction terms. IMPAS, infant mammalian penetration-aspiration score; LGG, left genioglossus; LTH, left thyrohyoid; RCT, right cricothyroid; RGG, right genioglossus.

Hypothesis 2: All Kinematic Variables and Some EMG Variables Predict IMPAS Score

All three kinematic variables, maximum hyoid excursion, posterior tongue ratio, and middle tongue ratio, predicted the occurrence of aspiration (Table 6). For each of these variables, the treatment (control vs. lesion) and the interaction between kinematics and lesion were significant. Hyoid excursion alone also predicted IMPAS score. The area under the ROC curve was relatively high for each. The odds ratio (OR) for the treatment variable was less than 0.2 for treatment, indicating that a lesion reduced the odds of having a normal (IMPAS = 1) swallow. The interaction (kinematics:treatment) ORs were all greater than 4.0, meaning that delayed time of maximum hyoid movement, greater posterior tongue ratio, and greater middle tongue ratio in controls were all more likely to produce a swallow with aspiration.

Table 6.

Significant logistic regression outcomes of airway protection score versus lesion status and oral kinematics, timing of muscle activation, and duration of muscle activation

Kinematic Variable	Factor Significance	Odds Ratio	ROC
PT ratio	PT ratio NS		0.7076
	Treatment SIG	<0.1
	PT ratio × treatment SIG	>200
MT ratio	MT ratio NS		0.6395
	Treatment SIG	<0.2
	MT ratio × treatment SIG	>4
Time of maximum hyoid elevation	Hyoid elevation SIG	>50	0.7887
	Treatment SIG	<.1
	Hyoid elevation × treatment SIG	>30

Muscle (Timing)	Factor Significance	Odds Ratio	ROC
Right genioglossus	Timing NS		0.7681
	Treatment SIG	>2
	Timing × treatment NS
Right cricothyroid	Timing NS		0.6825
	Treatment SIG	>2
	Timing × treatment NS
Right thyrohyoid	Timing NS		0.5656
	Treatment SIG	>12
	Timing × treatment SIG	<0.1
Left thyrohyoid	Timing NS		0.6218
	Treatment NS
	Timing × treatment SIG	>1
Right stylohyoid	Timing SIG	<1	0.7303
	Treatment SIG	>1
	Timing × treatment SIG	>1

Muscle (Duration)	Factor Significance	Odds Ratio	ROC
Right genioglossus	Duration SIG	>1	0.7681
	Treatment SIG	>2
	Duration × treatment NS
Left cricothyroid	Duration SIG	>1	0.821
	Treatment NS
	Duration × treatment NS
Right thyrohyoid	Duration SIG	>1	0.6802
	Treatment SIG	>6
	Duration × treatment SIG	<1
Left thyrohyoid	Duration SIG	>1	0.6205
	Treatment SIG	>2
	Duration × treatment SIG	<1
Right stylohyoid	Duration SIG	>1	0.6078
	Treatment SIG	>2
	Duration × treatment SIG	<1

Bold factor names indicate the coefficient for that factor in the logistic regression was significant at P < 0.05 (SIG, significant; NS, not significant). Odds ratios indicate magnitude and polarity of change in the odds of the outcome. An odds ratio greater than 1 indicates that the greater the value of the predictive factor, the greater the odds of an unsafe swallow. ROC, receiver operating characteristics; ROC area, measure of regression goodness of fit.

The EMG variables had a varied ability to predict IMPAS scores (Table 6). For timing of activity, several of the infrahyoid muscles, including thyrohyoid and cricothyroid, as well as stylohyoid, had significant treatment effects, as well as interactions. The ROC was high for all those regressions with significant ORs. Where there were significant effects, firing later increased the odds of IMPAS of 7.

For duration of activity, the right genioglossus was the only suprahyoid muscle with a significant result, with longer duration associated with increased odds of an IMPAS 7, and a high ROC. Longer duration of firing of the cricothyroid was also associated with increased odds of an IMPAS 7. Significant interactions with odds ratios of less than one were found for both sides of thyrohyoid and the right stylohyoid. In these muscles, a longer duration in lesions predicted IMPAS of 1 and a shorter firing duration predicted IMPAS of 7.

DISCUSSION

The Effects of Recurrent Laryngeal Nerve Lesion Differ among Physiological Systems

Swallow kinematics differed in lesion relative to nonlesion, for both safe and unsafe swallows. Prelesion, there was a difference in all three measures of kinematics between safe and unsafe swallows that disappeared after lesion. Yet, this interaction in response between lesion status and swallow safety was rarer in our measures of muscle activity. Where such interactions were significant, the post hoc results showed the same pattern as in the kinematic data: prelesion differences between safe and unsafe swallows disappeared postlesion. These results do not provide a clear pathway from muscle to movement to performance.

A possible objection to this interpretation is that there are more electromyographic variables in this study than kinematic ones, and thus, the study is biased against finding a strong pattern in the muscle activation data. However, in a broader study of kinematics after RLN lesion (Gross et al. 2018), half the variables showed an interaction between lesion and airway protection. In our electromyographic data, only one in seven muscles showed a similar interaction. A previous study (DeLozier et al. 2018) had shown that just under half the muscles changed either duration or timing in response to lesion alone. Our study highlights that those pervasive changes in muscle activation patterns are not closely linked with changes in airway protection outcomes. Kinematic changes postlesion seem to be more closely tied to airway protection outcomes than muscle activation.

An analogous lack of concordance between structure movement and muscle pattern exists in feeding among species of salamander (Lauder and Reilly 1996; Shaffer and Lauder 1985). This independent variation of kinematic and electromyographic data among closely related species suggests that different physiological levels of the feeding apparatus are evolutionary labile. The variability in our results might similarly suggest that even within a single species, the system has multiple potential pathways of muscle activity that produce successful swallowing. The functional redundancy in the hyoid musculature, for example, could mean that different combinations of timing and degree of muscle activity can produce the same type of hyoid movement (German et al. 2011). Adding to our sampling of hyoid muscles to include omohyoid, sternohyoid, and sternothyroid among the hyoid depressors and mylohyoid, geniohyoid, hyoglossus, and digastric among the hyoid elevators could address this question.

The previous work on specific components of swallowing biomechanics, both on this model species and clinical data on humans of all ages, demonstrates that kinematics and muscle pattern change with lesion (DeLozier et al. 2018; Gould et al. 2016; Haney et al. 2019; Pereira et al. 2006; Tsujimura et al. 2018). Furthermore, aspiration increases postlesion in all species where aspiration is possible (Gould et al. 2015; Périé et al. 1998) [owing to anatomical protection of the larynx, murine rodents do not aspirate (Haney et al. 2019)]. Yet, the goal to map changes in muscle function to specific kinematic differences, which in turn can be causally linked to swallow performance, is elusive. The documented complex relationships between muscle activation patterns and movements of primarily soft tissue structures with many degrees of freedom suggest that the effects do not propagate through levels of biological organization in a linear manner. One additional modality of data with the potential to help document these links is detailed data on changes in muscle lengths. Electromyographic signals, without such information, provide limited information on actual movements of oropharyngeal structures (Orsbon et al. 2018).

Predictive Value of Physiology for Performance

Kinematic variables showed a strong and consistent ability to predict changes in airway protection outcome. The significant interaction term between lesion and both tongue and hyoid variables, with a high odds ratio, indicates that changes in kinematics affect the probability of safe or unsafe swallow differently in control and lesion animals. Oropharyngeal movements across the musculoskeletal structures involved in swallowing respond in a coordinated fashion with regard to airway protection outcomes and lesion.

Muscle electrical activity shows a less consistent predictive pattern. For an infrahyoid muscle (thyrohyoid) and a suprahyoid muscle (stylohyoid), a change in timing and duration of activity predicts changes in airway protection differently pre- and postlesion. The odds ratios for duration are greater than one, but for timing, they are less than one, indicating that longer or shorter firing versus earlier or later firing affects airway protection differently. However, the only tongue muscle included in this analysis, genioglossus, does not have a significant interaction term, suggesting that although it could predict outcome, it was not able to distinguish between intact and lesioned swallows. Many muscles also showed no predictive power in this analysis, suggesting that, of the 20 or so muscles involved in swallowing, only a small subset are directly implicated in airway protection success or failure. An alternative interpretation to this result is that several small changes in the timing of muscle activity, below the power of this analysis, can produce a disordered swallow. The variable sample size of the EMG data relative to the kinematic data may make the EMG data sets less statistically powerful than the kinematics.

Understanding the Link between Muscle Electrical Activity, Kinematics, and Performance

The hierarchical relationship between motor pattern, kinematics, and airway protection in swallowing is nonlinear. Each level of the system has distinctive responses that are not simply a reflection of the level above. Thus, studies examining each level in isolation might lead to different conclusions, making identifying the mechanisms of airway protection failure difficult. An examination of kinematics alone suggests a strong interplay between recurrent laryngeal nerve lesion and airway protection (Gould et al. 2016). Yet, the results of this study show relatively weak effects on patterns of muscle activation, which reflect motor outputs from the brainstem. These differences hold across two different types of analysis. What is missing are more specific functional hypotheses that move beyond viewing swallow as a whole, but instead consider specific processes within the swallow that can be tied to the actions of specific muscles. Furthermore, identifying exactly how motor output from the brainstem controls functionally relevant variation in muscle activity and movement is necessary to understand neural control of swallowing performance and hypothesize potential feedforward and feedback mechanisms. The degree to which the system depends on feedforward (Criscimagna-Hemminger et al. 2010; Humbert et al. 2012) or direct feedback mechanisms will determine what relationships with airway protection we expect to see in the neural response following injury, as well as the relationship between motor output and kinematics. The RLN provides neither motor nor sensory innervation to any of the muscles measured in this study, or to the tongue or hyoid. Furthermore, the RLN senses aspiration once it has occurred, so its role in feedback during the swallow may be limited. However, the loss of sensory stimulus from the lower larynx could still be driving these changes by altering the regulation of brainstem swallowing centers through brainstem networks involving the respiration center, which interact with swallowing centers through reciprocal inhibition (McFarland and Lund 1993). In our study, the kinematics of safe and unsafe swallows are more similar postlesion, which suggests swallowing is more restricted in range of motion postlesion. This would be consistent with a strategy that uses little active feedback and relies instead on a fairly fixed motor pattern postlesion.

RLN lesion leads to changes in timing and duration of muscle activation in about half the muscles that have been measured (DeLozier et al. 2018). However, these results suggest that only a subset of these swallowing muscles are crucial to determining airway protection outcomes. Thus, subtle differences in the type of contraction, speed of contraction, and the timing and duration of activation of specific muscles within the swallow motor sequence may be important for neural control of airway protection (Thexton et al. 2009).

Limitations of the Study

Our study has several limitations that could affect interpretation of our results. We examined only a subset of possible kinematics and muscles involved in airway protection and swallowing. Sampling of additional kinematics might show a more nuanced picture than presented here and help identify what aspects of kinematics are key to successful airway protection. Similarly, sampling other muscles, including tongue muscles, the pharyngeal constrictors, and the intrinsic laryngeal muscles, would improve our understanding of the relationship between neuromotor output and airway protection failure. Finally, although airway protection is a key aspect of swallowing performance, it is only one way of measuring effectiveness of swallowing. Bolus size and swallow time, parameters that relate to the efficiency of nutrient acquisition, also reflect swallow performance, particularly for juveniles. Analyzing how feeding efficiency in swallowing is related to kinematics and muscle function may reveal potential performance trade-offs with airway protection (Gould et al. 2017; Mayerl et al. 2020).

Finally, our study does not explicitly consider the role of respiration in the control of swallowing, or how RLN lesion might affect it. Swallowing and respiration are antagonistic functions, which use the same anatomical substrates, and thus are tightly coordinated (Ballester et al. 2018; McFarland and Lund 1993). Increasing respiratory drive through hypercapnia or other ways can change or entirely suppress swallowing rates (Dick et al. 1993; Nishino et al. 1986; Timms et al. 1993). The recurrent laryngeal nerve is important in upper airway respiratory responses such as cough reflex (Canning 2006). Thus, it is possible that unilateral recurrent laryngeal nerve lesion leads to increased respiratory drive, which decreases swallow respiration coordination, leading to the changes we see here. Recurrent laryngeal nerve lesion has been shown to affect swallow respiration coordination maturation in infant pigs (Ballester et al. 2018) although the impact on swallow safety is limited (Stricklen et al. 2020).

Conclusion

The effect of unilateral recurrent laryngeal nerve lesion on swallowing manifests differently in kinematic and electromyographic data from the same animals, in muscles and sensory regions not innervated by the recurrent laryngeal nerve. Changes in kinematics are more closely tied to postlesion differences in airway protection than changes in muscle activation patterns. Kinematics predict airway protection outcomes more consistently than muscle activation data. Thus, a better understanding of how muscle activity relates to movement of structures and ultimately of the bolus is needed to find the mechanistic links between recurrent laryngeal nerve lesion and airway protection failure, and identify targets of neural control. When measuring the effects of an intervention on behavioral outputs, it is important to consider what may be occurring in the parts of the system not explicitly being recorded, as they have effects on outcomes.

GRANTS

This work was funded by National Institutes of Health Grant DC-9880 (to R. Z. German).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

F.D.H.G., J.O., and R.Z.G. conceived and designed research; F.D.H.G., A.R.L., J.O., and R.Z.G. performed experiments; F.D.H.G. and R.Z.G. analyzed data; F.D.H.G., A.R.L., C.M., and R.Z.G. interpreted results of experiments; F.D.H.G. prepared figures; F.D.H.G. drafted manuscript; F.D.H.G., A.R.L., C.M., J.O., and R.Z.G. edited and revised manuscript; F.D.H.G., A.R.L., C.M., J.O., and R.Z.G. approved final version of manuscript.