Tongue electrical impedance myography correlates with functional, neurophysiologic, and clinical outcome measures in long-term oropharyngeal cancer survivors with and without hypoglossal neuropathy: an exploratory study

Abstract

Background:

This pilot study analyzed correlations between tongue electrical impedance myography (EIM), standard tongue electromyography (EMG), and tongue functional measures in N=4 long-term oropharyngeal cancer (OPC) survivors.

Methods:

Patients were screened for a supportive care trial (NCT04151082). Hypoglossal nerve function was evaluated with genioglossus needle EMG, functional measures with the Iowa oral performance instrument (IOPI), and multi-frequency tissue composition with tongue EIM.

Results:

Tongue EIM conductivity was higher for patients with EMG-confirmed cranial nerve XII neuropathy than those without (p = 0.005) and in patients with mild versus normal EMG reinnervation ratings (16 kHz EIM: p = 0.051). Tongue EIM correlated with IOPI strength measurements (e. g. anterior maximum isometric lingual strength: r² = 0.62, p = 0.020).

Conclusions:

Tongue EIM measures related to tongue strength and the presence of XII neuropathy. Non-invasive tongue EIM may be a convenient adjunctive biomarker to assess tongue health in OPC survivors.

Introduction

Human papillomavirus (HPV)-associated oropharyngeal cancer (OPC) incidence has risen significantly over the last two decades and is projected to rise until the mid-2030s.^1–4 Almost half of head and neck cancer (HNC) cases are now HPV-driven OPC.⁵ Even with current vaccination rates, the epidemic rise in the incidence of HPV-associated OPC is not expected to decline until at least 2045.² HPV has an excellent prognosis for long-term survival.⁶ Despite being highly curable, late toxicities of radiation therapy (RT) treatments can adversely impact survivors’ quality of life (QOL) and function.⁷ Normal tissue injury during RT or surgery can induce lower cranial neuropathy (LCNP), most commonly affecting the hypoglossal (XII) cranial nerves with glossopharyngeal (IX), vagus (X), and accessory (XI) nerves also at risk.^6,7 Damage to these nerves can impair normal swallowing, speech, and shoulder function.^8,9

Cranial neuropathy independently associates with excess symptom burden in OPC survivors, with the most considerable effect on critical functions of eating and speaking.⁶ Late feeding tube insertion, tracheostomy, and pneumonia events almost exclusively occur in survivors who suffer late cranial neuropathy.¹⁰ Previous studies specifically identify LCNP as a significant contributor to late radiation-associated dysphagia (late-RAD).¹⁰ Poor tongue function due to late-RAD can induce adverse effects such as mucus problems, choking, and reduced swallowing-related QOL. Survivors are 2.7 times more likely to develop life-threatening aspiration pneumonia (airway entry of liquids or food) than non-cancer controls, which confers a 42% increased mortality risk among survivors.¹¹ Early identification of LCNP and, thus, RAD-at-risk individuals is the first step toward targeting more effective therapies or optimized treatment algorithms to intervene more effectively before irreversible features of RT injury manifest.

Radiotherapy-related toxicities are most commonly assessed through functional outcome measurement. Symptoms are typically monitored using patient-reported outcome (PRO) instruments such as the MD Anderson symptom inventory head and neck cancer module (MDASI-HN) symptom scores and clinician-graded methods such as the performance status scale for head and neck cancer (PSS-HN) assessment, videofluoroscopic modified swallowing study (MBS),¹² and strength via the Iowa oral performance instrument (IOPI). These measures provide insight into tongue function once symptoms or observable functional changes manifest, but they may not fully or conveniently monitor tongue health between RT completion and the development of impairment. Gold-standard needle electromyography (EMG) provides neurophysiologic data to classify lower cranial nerve health. Despite the clear value, limitations of the EMG, including the subjective interpretation and invasive nature, do not allow it to follow the progression and response to therapy in OPC survivors at large.^13,14

Tongue electrical impedance myography (EIM) has recently been introduced as a potential bedside tool for monitoring tongue health in bulbar disease states.^15–17 EIM measures muscle volume electrical conductivity properties to detect alterations in tissue composition (e.g., deposition of fatty infiltration and connective tissue) and structure (e.g., changes of muscle fiber type and myofiber atrophy).^18–20 Specifically, tongue EIM has shown promise in evaluating neuromuscular disorders affecting the tongue,²¹ including amyotrophic lateral sclerosis.^22–24 Compared to EMG, performing tongue EIM measurements is non-invasive and quick, thus has the potential to serve as an adjunctive quantitative tool to follow progression and monitoring response to therapy in OPC survivors.

Our hypothesis is that tongue EIM in OPC correlates with neurophysiological and functional outcomes, and patients with LCNP will exhibit different tongue EIM values from those previously reported in healthy subjects in Luo et al.¹⁶ Because pathological states are mediated by fibrotic tissue and associated atrophy, we expect higher tongue conductivity values in relation to pathological EMG and IOPI outcomes, which would be consistent with previously reported tongue EIM values.¹⁸

Materials and methods

Subject information

Post hoc analysis of exploratory tongue EIM was performed among N=4 consecutively enrolling participants on NCT04151082. Data collection for this study occurred between November 5, 2021, and October 25, 2022. NCT04151082²⁵ is an ongoing clinical trial recruiting long-term OPC survivors who develop late radiation-associated LCNP. Patients enrolled in this pilot study are administered prednisone as part of the parent trial, NCT04151082, which focuses directly on evaluating the potential of immunosuppressive prednisone to provide symptomatic improvement for OPC survivors with LCNP. Eligible patients are disease-free adults (18 or older) who completed RT for at least two years before study enrollment and consented to return for assessment post-steroid therapy. They exhibit one or more of the following symptoms: lingual deviation, fasciculations, atrophy, and EMG-confirmed LCNP. Patients with uncontrolled diabetes or hypertension, a history of psychosis, untreated refractory obstructive pharyngoesophageal stricture, a history of bipolar disorder, and pregnant women were excluded from the study. Participants signed an informed consent document to participate in this study approved by the MD Anderson Cancer Center Institutional Review Board (#2019–0207).

Study procedure

Participants were evaluated with functional and clinical outcome measures in three data-gathering sessions. Photos were taken of the patients’ tongues during each session. Baseline measures were taken upon study enrollment. Patients then received prednisone PO (or by feeding tube) QD on days 1–5 and then tapered off over two weeks.²⁶ Functional and physiological outcomes were collected approximately two weeks post-taper. The third measurement session occurred 6–10 weeks after post-taper measurements. The longitudinal timeframe for each patient is shown in Table 1.

Table 1:

Longitudinal timeframe for functional and physiological outcomes reporting

Patient	Baseline	Post-Taper	6–10 Weeks Post-Taper
0012	11/5/2021	02/10/2022	03/24/2022
0015	03/10/2022	05/19/2022	06/23/2022
0016	04/11/2022	06/02/2022	10/25/2022
0017	05/18/2022	06/30/2022	08/04/2022

Study data collection occurred between November 5, 2021, and October 25, 2022. Baseline measures were taken upon study enrollment. The second session occurred approximately four weeks post-taper. The third measurement session occurred 6–10 weeks after post-taper measurements.

Needle electromyography

Intramuscular EMG assessed the spontaneous insertional activity and voluntary motor unit potential (MUP) in the tongue via a concentric needle electrode inserted via the floor of the mouth into the genioglossus muscle of the tongue. An electromyographer (K.W.) classified XII nerve function by scoring at-rest muscle activity based on a 4-point denervation potentials grade, with 0 being none and 4 being severe degree of fibrillations and positive sharp waves signifying denervation.^13,14 To test voluntary contraction, the patient was asked to protrude the tongue, and EMG was graded on a 4-point reinnervation potentials grade, with 1 being normal and 4 being severe reinnervation and abnormal morphology in the motor unit (wider duration, larger amplitude, polyphasic activity).^27,28

Iowa oral performance instrument

IOPI is a digital manometer that was used to evaluate tongue strength.²⁹ Maximum isometric lingual strength (MILS) and maximum swallowing lingual strength (MSLS) were measured to assess the maximum pressure generated by the tongue. A small rubber bulb was attached to a pressure transducer. The pressure exerted by the tongue against the bulb positioned against the hard palate was measured in kilopascals (kPa). MSLS and MILS values were recorded with the bulb positioned in the anterior and posterior regions of the tongue. Maximum strength was determined by calculating the maximum of three peak force generation efforts within 10% variance.³⁰ Participants were encouraged to rest between trials for a minimum of 5 breath cycles.

Lingual range of motion

Lingual range-of-motion (LROM) is measured by observing protrusion, left lateralization, right lateralization, and elevation, and scored according to the scale defined by Lazarus et al.³¹ The subject’s total score will range from 100 (normal) to 0 (totally impaired).

Performance status scale for head and neck cancer patients

The PSS-HN is a questionnaire-guided semi-structured interview consisting of three questions: normalcy of diet, public eating, and understandability of speech.³² Single-item scores were reported for each of the three domains (Diet, Speech, Eating) with a range of 0 (non-functional) to 100 (full functioning).

MD Anderson symptom inventory for head and neck cancer

The MDASI-HN is a patient-reported outcome questionnaire designed to measure the severity or burden of systemic and head- and neck-specific symptoms and their interference with or effect on patients’ daily functioning. This 28-item multi-symptom inventory is rated according to the last 24 hours at their worst, with 0 being symptoms not present and 10 being as bad as the patient can imagine. The trial’s primary endpoint from the MDASI-HN is the patient’s mean score for the following five survey items: i, difficulty chewing/swallowing; ii, difficulty speech/voice; iii, choking; iv, dry mouth; and v, mucus. Diminished lingual strength and range of motion should elevate the symptom burden caused by these effects.^6,33

Modified barium swallow study

The MBS was used to identify oropharyngeal swallow physiology, aspiration, and pharyngeal residue. As a gold-standard measure of dysphagia in clinical practice, MBS video recordings were scored by a trained speech pathologist blinded to clinical history for two measures: i, dynamic imaging grade of swallowing toxicity (DIGEST)³⁴ and ii, the penetration-aspiration scale.³⁵ DIGEST was the primary endpoint from MBS. The MBS protocol was standardized to include two trials each of 5mL, 10mL, and sips from a cup of Varibar thin liquid, Varibar barium pudding, and a cracker dipped in Varibar barium pudding. The study team developed and internally validated DIGEST as an MBS-derived dysphagia grade compatible with the National Cancer Institute’s Common Terminology Criteria for Adverse Events (1 = mild, 2 = moderate, 3 = severe, 4 = life-threatening).^12,36

Tongue electrical impedance myography correlatives

Tongue EIM data for each patient was collected via the user tongue electronic system (UTES).^16,37 Data was recorded as five repeated measures of tongue conductivity (Siemens/meter; S/m) at six different frequencies of alternating electrical current. The tongue EIM was collected using a custom tongue blade called the user tongue array (UTA), which is 10 cm long and 1.8 cm wide (Figure 1A). The tongue contact of the UTA probe consists of 16 surface electrodes arranged in two concentric circles. The outer circle has a radius of 0.75 cm, and the inner circle has a radius of 0.4 cm (Figure 1B). The other end of the UTA probe connects to the tongue EIM UTES recording device.

Figure 1:

The user tongue electronic system (UTES). A: The UTES tongue probe measures 10 cm × 1.8 cm. B: The electrode array on the end of the probe consists of two concentric circles of electrodes. The outer circle has a radius of 0.75 cm, and the inner circle has a radius of 0.4 cm.

Our investigational UTES device connects to a custom smartphone application via WiFi. The user follows on-screen instructions to enter operator and de-identified subject information. The tongue EIM probe is placed on the center of the patient’s tongue and initiates a measurement from the smartphone. The UTES device measures tongue EIM at 8, 16, 32, 64, 128, and 256 kilohertz (kHz). The device automatically filters artefactual measured data and sends tongue EIM information to the smartphone. Data is then directly emailed from the app upon completion of the measurement session. One tongue EIM measurement takes approximately 30 seconds to complete; participants completed a maximum of five measurements.

Data analysis

Descriptive and univariate analyses were performed. For tongue EIM, mean conductivity values were calculated for each frequency and measurement session. Statistical differences between two samples of continuous data were analyzed using the non-parametric t-test using Prism software version 9.5.1 (GraphPad Software Inc, San Diego, CA). Statistical differences between tongue EIM values for each discrete EMG rating were analyzed using one-way analysis of variance (ANOVA) with Tukey correction for multiple comparisons. Participants’ IOPI and tongue EIM measurements were divided by baseline value, and linear regression was used to estimate the strength of correlations between IOPI and tongue EIM. Outliers were detected and eliminated using GraphPad’s regression and outlier removal.³⁸ A two-tailed α-level of 0.05 was considered significant, and a 0.10 threshold was considered marginally significant for these exploratory hypothesis-generating analyses. See supplementary figures for tests meeting significance at α = 0.10.

Results

Sample characteristics

Four disease-free HNC survivors with late-RAD undergoing screening procedures (including EMG, IOPI, and functional testing) for enrollment into a supportive care trial for late radiation-associated cranial neuropathy were included. All patients were previously treated with RT (between 2 to 25 years prior) for primary tumors of the oropharynx and nasopharynx. Three oropharyngeal cancer patients (75%) exhibited cranial nerve XII neuropathy by needle EMG. The remaining patient had no evidence of XII neuropathy on needle EMG; however, they had bilateral vocal fold paralysis that was presumed to be a late effect of RT by excluding other sources of injury. Table 2 summarizes the characteristics of the study sample as recorded at the baseline measurement session.

Table 2:

Late radiation-associated XII neuropathy cases

Characteristic	Value
Total Sample No.	4
Sex, no. (%)
Men	3 (75%)
Women	1 (25%)
Age, median (range)	62 (56–71)
Years Between RT and Steroid Trial, median (range)	7.5 (2.33–25)
Tumor site, no. (%)
Oropharynx	3 (75%)
Nasopharynx	1 (25%)
Nerve XII Neuropathy (per needle EMG)
Yes	3 (75%)
No	1 (25%)
Tongue strength (median, IQR)
MILS – anterior	29.5 (31)
MILS – posterior	21 (26.5)
MSLS – anterior	19.5 (21.25)
MSLS – posterior	10.5 (10.75)
Lingual Range of Motion (LROM)
LROM 100	2 (50%)
LROM <100	2 (50%)
PSS-HN
Normalcy of Diet
• Regular diet (liquid wash) (90)	1 (25%)
• Soft chewable (50)	1 (25%)
• Soft foods requiring no chewing (40)	1 (25%)
• Non-oral feeding (tube-fed) (0)	1 (25%)
Eating in Public
• No restriction of place, food, or companion (100)	1 (25%)
• No restriction of place, but restricts diet in public (75)	3 (75%)
Understandability of Speech
• Always understandable (100)	2 (50%)
• Understandable most of the time; occasional repetition necessary (75)	2 (50%)
MDASI-HN
Difficulty chewing/swallow
• 3	1 (25%)
• 4	1 (25%)
• 7	1 (25%)
• 9	1 (25%)
Difficulty speech/voice
• 6	1 (25%)
• 7	1 (25%)
• 9	2 (50%)
Choking
• 3	1 (25%)
• 6	2 (50%)
• 9	1 (25%)
DIGEST
Safety
• 0 - WNL	1 (25%)
• 2 – Moderate	1 (25%)
• 3 – Severe	1 (25%)
• N/A - Unavailable	1 (25%)
Efficiency
• 2 – Moderate	1 (25%)
• 3 – Severe	2 (50%)
• N/A - Unavailable	1 (25%)

Abbreviations: RT, radiotherapy; EMG, electromyography; IQR, interquartile range; MILS, maximum isometric lingual strength; MSLS, maximum swallowing lingual strength; LROM, lingual range of motion; PSS-HN, Performance Status Scale for Head and Neck Cancer Patients; MDASI-HN, MD Anderson Symptom Inventory Head and Neck Cancer Module; DIGEST, Dynamic Imaging Grade of Swallowing Toxicity.

Patient clinical outcomes

Figure 2 shows images of tongue protrusion effort for each patient. Photos were taken straight on and to the right and left of the face using grids to standardize alignment. Patients 0012 and 0016 exhibited LROM < 100 and the most restrictive PSS-HN scores, where patient 0012 diet was limited to soft foods, and 0016 had a tracheostomy and was tube-fed. Two participants, 0015 and 0017, had normal tongue range of motion per LROM = 100 and less diet restriction per PSS-HN. Participants with LROM < 100 reported significantly worse MDASI-HN scores at 6–10 weeks post-taper than those with LROM = 100 (LROM < 100: 8.33 ± 1.02, 95% CI, 1.71 to 6.96 vs LROM = 100: 4.33 ± 0.67, 95% CI, 6.62 to 10.05; p = 0.008). Participants with XII neuropathy reported significantly worse MDASI-HN scores 6–10 weeks post-taper than the patient without EMG-detected XII neuropathy (XII neuropathy: 7.33 ± 0.71, 95% CI, 5.70 to 8.96 vs. no XII neuropathy: 3.33 ± 1.86, 95% CI, −4.65 to 11.32; p = 0.030).

Figure 2:

Baseline patient tongue protrusion images. Patients 0015 and 0017 exhibited no range of motion limits. Patient 0012 has a mildly limited range of motion. Patient 0016 exhibits a severely limited range of motion.

Patient 0016 outcomes

Patient 0016 experienced the most severe impact to QOL, as indicated by several metrics. Throughout the study, patient 0016 DIGEST scores per MBS were graded between severe (3) and life-threatening (4) pharyngeal dysphagia. All other participants’ DIGEST scores ranged between mild (1) to moderate (2). Through patient-reported outcomes, patient 0016 reported significantly worse MDASI-HN scores than the rest of the cohort (1.70, p = 0.045). Clinically, patient 0016 showed significantly lower MILS tongue strength than the rest of the cohort (−31.22 kPa, p = 0.001).

Tongue electrical impedance myography analysis by XII neuropathy classification

Tongue EIM at 8 and 128 kHz, shown in Figure 3A and 3B respectively, was higher for patients with XII neuropathy (8 kHz: mean difference, 0.290 S/m; 95% CI, 0.113 to 0.467 S/m; p = 0.005), (128 kHz: mean difference, 1.109 S/m; 95% CI, −0.012 to 2.24 S/m, p = 0.053). No significant differences were found at other measured frequencies.

Figure 3:

Tongue EIM. The total 8 (A) and 128 (B) kHz tongue EIM samples are divided between EIM measures among patients with no XII neuropathy (n = 3) and XII neuropathy per needle EMG (n = 6 and 8, respectively). Patients with XII neuropathy exhibit higher conductivity (A: mean difference, 0.290 S/m; 95% CI, 0.113 to 0.467 S/m; p = 0.005) and (B: mean difference, 1.109 S/m; 95% CI, −0.012 to 2.24 S/m; p = 0.053).

Tongue electrical impedance myography versus electromyography reinnervation grading

Figure 4 shows the relationship between 16 (A) and 128 (B) kHz tongue conductivity values and their corresponding genioglossus reinnervation EMG rating. Analysis shows a marginally significant increase in tongue conductivity when the reinnervation EMG rating increases from normal EMG (rating 1) to mild reinnervation EMG (rating 2) for 16 kHz (0.33 S/m, 95% CI: 0.01 to 0.67 S/m, p = 0.053, n = 8) and 128 kHz (0.32 S/m, 95% CI: 0.00 to 0.64 S/m, p = 0.051, n = 9). Figure S1 shows considering significance α = 0.10, 8 kHz tongue EIM also significantly increases between EMG rating 1 to 2 (0.29 S/m, 90% CI: 0.03 to 0.54 S/m, p = 0.069, n = 8). We found no significant difference in EIM between denervation EMG categories.

Figure 4:

Comparison between needle EMG and tongue EIM. Marginally significant differences in conductivity at 16 (A) and 128 (B) kHz between EMG reinnervation rating 1 vs. 2 (mean difference, 0.33 S/m; 95% CI, 0.01 to 0.67 S/m; p = 0.053; n = 8) and (mean difference, 0.32 S/m; 95% CI, 0.00 to 0.64 S/m; p = 0.051; n = 9), respectively. The sample mean is depicted by the bars.

Tongue electrical impedance myography versus Iowa oral performance instrument

Figure 5A shows that tongue EIM correlates with anterior and posterior MILS measures (8 kHz vs. anterior MILS: r² = 0.62, p = 0.020; 16 kHz anterior MILS: r² = 0.68, p = 0.012; 16 kHz vs. posterior MILS: r² = 0.50, p = 0.048). Figure 5B shows that tongue EIM correlates with anterior and posterior MSLS measures (8 kHz vs. anterior MSLS: r² = 0.51, p = 0.047; 16 kHz vs. anterior MSLS: r² = 0.71, p = 0.008; 8 kHz vs. posterior MSLS: r² = 0.54, p = 0.037; 32 kHz vs. posterior MSLS: r² = 0.58, p = 0.028). Figure S2A shows that tongue EIM correlates with anterior and posterior MILS measures (32 kHz vs. anterior MILS: r² = 0.39, p = 0.0996; 32 kHz vs. posterior MILS: r² = 0.49, p = 0.053; 128 kHz vs. anterior MILS: r² = 0.35, p = 0.091). Figure S2B shows that tongue EIM correlates with anterior and posterior MSLS measures (32 kHz vs. anterior MSLS: r² = 0.46, p = 0.063; 32 kHz vs. posterior MSLS: r² = 0.40, p = 0.094). All samples are of size n = 8, and supplementary figures considered α = 0.10.

Figure 5:

Correlation of tongue EIM and IOPI. Value-to-baseline ratios of 8, 16, and 32 kHz tongue EIM significantly correlate with value-to-baseline ratios of MILS (A) and MSLS (B). (A: 8 kHz vs. anterior MILS: r² = 0.62, p = 0.020; 16 kHz anterior MILS: r² = 0.68, p = 0.012; 16 kHz vs. posterior MILS: r² = 0.50, p = 0.048). (B: 8 kHz vs. anterior MSLS: r² = 0.51, p = 0.047; 16 kHz vs. anterior MSLS: r² = 0.71, p = 0.008; 8 kHz vs. posterior MSLS: r² = 0.54, p = 0.037; 32 kHz vs. posterior MSLS: r² = 0.58, p = 0.028). All samples are of size n = 8, and some plots have overlapping points.

Discussion

This pilot study compared experimental tongue EIM (electro-histology), needle EMG (neurophysiology), and IOPI (functional) measurements in a small subset of trial participants representing long-term OPC survivors presenting with clinical signs of late radiation-associated cranial neuropathy and late-RAD. We compared tongue EIM data collected longitudinally during the supportive care trial to clinical EMG and IOPI performance measures. Tongue EIM conductivity was higher for individuals who exhibit XII neuropathy per gold-standard EMG than the participant who did not. Patients who exhibited mild reinnervation grades representing mild chronic nerve injury and abnormal motor unit morphology also had higher tongue conductivity than those who did not exhibit these signs of chronic hypoglossal neuropathy. Tongue EIM conductivity was also shown to decrease as IOPI tongue performance improved in both anterior isometric strength and posterior swallowing strength.

The four patients in this small cohort exhibited physiological symptoms along the full severity continuum, from mild to life-threatening. Patient 0016 was enrolled in this trial 25 years post-RT, and their symptom severity unsurprisingly poses the most significant detriment to QOL and function. Notably, patient 0016 also had the highest tongue conductivity, representing the worst degree of compositional changes. Patient 0016’s severe QOL detriment highlights the urgent need for earlier neuropathy detection, which might guide and motivate the prescription of tongue strength and swallowing exercises to prevent further atrophy through disuse or contralateral tongue compensation.^39–41 It is possible that pre-symptomatic, quantitative biomarkers of tongue health might motivate clinicians and their patients to engage in proactive rehabilitation efforts against late effects. Simple, non-invasive testing would be mandatory to achieve this clinical goal.

Tongue composition and structure changes due to co-existing normal tissue injury, including fibrosis and neuropathy post-RT, are both reported in OPC survivors who develop late-RAD.^42,43 Previous research has found that approximately 47% of HNC patients develop moderate to severe fibrosis within 12 months of RT.⁴⁴ Fibrosis is thought to compress peripheral nerve tracts, thereby contributing to the denervation of critical swallowing muscles, but earlier inflammatory or vascular changes may underlie nerve dysfunction reported to occur in 10% of survivors over the first 2 decades of survivorship.^45–47 Our statistical analysis shows that tongue EIM at 8 kHz conductivity is significantly higher for patients with XII neuropathy than for those without (p = 0.005), thereby suggesting that tongue EIM might be useful to detect compositional tongue changes that relate to eventual XII neuropathy as likely mediated through the presence of lingual fibrosis. Of note, the resultant average tongue conductivity for patients with normal EMG grade and no XII neuropathy (128 kHz reinnervation grade 1: 0.4112 S/m; 128 kHz no XII neuropathy: 0.5129 S/m) is within the range of healthy values reported by Luo et al.¹⁶ Along the continuum of OPC survivorship after RT, the tongue compositional changes are complex, and in this pilot dataset spanning 2 to 25 years post-treatment, the elevated EIM could reflect a host of compositional changes including tongue fibrosis alongside fatty infiltration and muscle atrophy after years after onset of clinically evident XII neuropathy.^48,49 Tongue EIM is expected to be sensitive to changes in tissue composition, where different tissue types are known to exhibit unique electrical signatures.⁵⁰ Fibrotic tissue has been reported to exhibit higher conductivity values than healthy muscle, whereas infiltrated fat has reportedly lower conductivity values over the frequency range measured here.^50,51 Whereas EIM measures electro-histological features of the tongue to detect abnormal muscle composition, our findings also suggest relevance to EMG values. These findings support the potential role of our approach to surface EMG (sEMG) tongue measurement. Combining non-invasive tongue EIM and sEMG neurophysiological signals might provide investigators and clinicians another route to discriminate muscle fibrosis from cranial neuropathy in survivors suffering late effects.

Post-RT fibrosis and XII neuropathy significantly impact patient functional status.⁴² Our results demonstrate that, as tongue strength increases, EIM-measured tongue conductivity decreases (r² > 0.5, p < 0.05). The relationships described by this comparison suggest that an increase in swallowing strength and isometric strength may coincide with more normal tongue muscle composition and decreased tongue electrical conductivity associated with the presence of healthy myofibers. Our findings are consistent with previous work studying tongue EIM application in amyotrophic lateral sclerosis,^18,52 where tongue EIM was also found to correlate with tongue strength due to changes to the composition of the tongue to more electrically conductive tissue types.¹⁸

This pilot study has several limitations. Extreme atrophy is commonly observed in patients in this setting, and such anatomical changes influenced tongue strength testing. In our ongoing follow-up study we are now also performing lateralized tongue EIM measurements and added tongue elevation tasks to more directly quantify the hemi-atrophy of the tongue. We recruited a small patient cohort, all of whom were symptomatic with late effects of RT, which limits the generalizability of our findings. Our pilot analysis groups three measurements over time into one sample due to the relative brevity of the study timeline. In the future, we aim to expand our preliminary work to investigate tongue EIM trends over long-term post-RT toxicity progression. Future work will also focus on measuring a larger sample of post-RT patients, including a healthy control group, to generate normative healthy tongue EIM values within this patient population. These data may prove insightful when gathered in parallel with pre-habilitation efforts or active rehabilitation among patients with dysphagia.^40,42 Finally, we plan to implement a machine learning algorithm for patient risk stratification using future longitudinal tongue EIM data.

Cranial XII neuropathy is a prevalent late toxicity of RT for OPC survivors, leading to dysphagia with an increased risk of feeding tube insertion, tracheostomy, and aspiration pneumonia. Current clinical gold standard symptom monitoring methods lack specificity to physical changes in tongue health necessary to monitor normal tissue recovery and injury after RT. This study is the first to investigate tongue EIM in this patient population. The findings of this pilot study suggest that experimental tongue EIM data may serve as an informative, non-invasive, quantitative biomarker for tongue health (both neurophysiology and strength) in pharyngeal cancer survivors after RT. Ongoing and future work is currently focused on establishing the feasibility of non-invasive tongue EIM as well as sEMG as biomarkers of XII neuropathy through cross-sectional and longitudinal measurements in a larger cohort of OPC survivors and healthy participants.

Supplementary Material

Fig S1

Figure S1: Comparison between needle EMG and tongue EIM when α = 0.10. The difference in conductivity at 8 kHz between reinnervation rating 1 vs. 2 (mean difference, 0.29 S/m; 90% CI, 0.03 to 0.54 S/m; p = 0.069; n = 8).

Fig S2

Figure S2: Correlation of tongue EIM and IOPI when α = 0.10. Value-to-baseline ratios of 32 and 128 kHz tongue EIM significantly correlate with value-to-baseline ratios of MILS (A) and MSLS (B). (A: 32 kHz vs. anterior MILS: r² = 0.39, p = 0.0996; 32 kHz vs. posterior MILS: r² = 0.49, p = 0.053; 128 kHz vs. anterior MILS: r² = 0.35, p = 0.091). (B: 32 kHz vs. anterior MSLS: r² = 0.46, p = 0.063; 32 kHz vs. posterior MSLS: r² = 0.40, p = 0.094). All samples are of size n = 8, and some plots have overlapping points.

Abbreviations:

CI

Confidence Interval