Does notching along the nasal velar surface during nasopharyngoscopy predict discontinuity of the underlying levator veli palatini muscle?

Abstract

Objective:

To determine the sensitivity and specificity of velar notching seen on nasopharyngoscopy for levator veli palatini (LVP) muscle discontinuity and anterior positioning.

Design:

Nasopharyngoscopy and MRI of the velopharynx were performed on patients with VPI as part of their routine clinical care. Two speech-language pathologists independently evaluated nasopharyngoscopy studies for the presence or absence of velar notching. MRI was used to evaluate LVP muscle cohesiveness and position relative to the posterior hard palate. To determine the accuracy of velar notching for detecting LVP muscle discontinuity, sensitivity, specificity, and positive predictive value (PPV) were calculated.

Setting:

A craniofacial clinic at a large metropolitan hospital.

Participants:

Thirty-seven patients who presented with hypernasality and/or audible nasal emission on speech evaluation and completed nasopharyngoscopy and velopharyngeal MRI study as part of their preoperative clinical evaluation.

Results:

Among patients with partial or total LVP dehiscence on MRI, presence of a notch accurately identified discontinuity in the LVP 43% (95% CI 22–66%) of the time. In contrast, the absence of a notch accurately indicated LVP continuity 81% (95% CI 54–96%) of the time. The PPV for the presence of notching to identify a discontinuous LVP was 78% (95% CI 49–91%). The distance from the posterior edge of the hard palate to the LVP, known as effective velar length, was similar in patients with and without notching (median 9.8 mm vs 10.5 mm, p =1.00).

Conclusion:

The observation of a velar notch on nasopharyngoscopy is not an accurate predictor of LVP muscle dehiscence or anterior positioning.

Introduction

It is estimated that up to 40% of children with a repaired cleft palate will develop velopharyngeal insufficiency (VPI).^1–4 The decision for surgical management is based on a perceptual speech assessment and imaging of the velopharynx. The goal of velopharyngeal imaging is to provide direct visualization of velopharyngeal function during speech to confirm presence of a velopharyngeal gap as the cause of hypernasality and to assist in treatment decision making. Reliability and validity of imaging assessments is paramount, as results directly inform the treatment plan.

Nasopharyngoscopy is the most common clinical imaging tool used for VPI surgical planning. Prior surveys indicate that 59% of cleft teams across the United States use nasopharyngoscopy as the primary imaging tool.⁵ Nasopharyngoscopy enables an assessment of velar excursion, lateral pharyngeal wall movement, closure pattern, and velopharyngeal gap size and location during speech production.⁶

Sie and Chen further recommend the use of nasopharyngoscopy for all patients with VPI to determine the status of the LVP musculature. The authors specify a notch or groove seen on the nasal velar surface during nasopharyngoscopy to be indicative of sagittally-oriented and anteriorly positioned LVP muscle bundles.^7,8 The LVP fibers originate at the base of the skull and the posterior aspect of the Eustachian tube.^9,10 In non-cleft anatomy the two muscle bundles descend anteriorly, medially, and inferiorly to insert into the body of the velum where they converge and join with the opposing levator bundle forming a cohesive sling. This normal crossing of the LVP fibers create a transverse arrangement of the intravelar LVP muscle fibers.

In an overt cleft, the LVP fibers do not cross the velar midline due to the tissue deficit and instead follow an anterior-posterior or sagittal orientation. While the goal of primary palatoplasty is to create this normal transverse LVP sling arrangement, in some cases of VPI, the muscle remains or migrates to its prior sagittal orientation. Many teams have adopted these recommendations and similarly use velar notching as an indicator of LVP status and relative position.^8,11–15 Furthermore, teams routinely describe the presence versus absence of velar notching on nasopharyngoscopy as a diagnostic criterion for recommendation of palatoplasty (i.e., Furlow Z-Plasty) versus pharyngoplasty (i.e. pharyngeal flap or sphincter) for management of VPI.

Despite the wide use of nasal velar notching as an indicator of LVP pathology, the accuracy of this finding has not been investigated. The purpose of this study was to determine the sensitivity and specificity of velar notching identified during nasopharyngoscopy for discontinuity and anterior positioning of the LVP muscle fibers on velopharyngeal MRI.

Methods

Subjects:

Following receipt of institutional review board approval, a retrospective review was performed of patients undergoing MRI of the velopharynx and nasopharyngoscopy as part of their preoperative VPI evaluation at Phoenix Children’s (Phoenix, Arizona) between July 2018 and June 2021. Patients meeting all of the following criteria were included: (1) presence of hypernasality and/or audible nasal emission on speech evaluation, (2) successful completion of preoperative MRI and (3) successful completion of preoperative nasopharyngoscopy which was videorecorded and available for review. There were no exclusion criteria.

Standardized speech evaluation:

Prior to MRI and nasopharyngoscopy, all patients were seen by a cleft team speech-language pathologist for a clinical visit and completed a standardized speech evaluation. The perceptual speech evaluation included an audio or video recorded speech sample with conversation, counting, and a phrase or sentence repetition task. The Cleft Audit Protocol for Speech-Augmented American Americleft Modification (CAPS-A-AM) rating scale was used to categorically rate the severity of speech parameters related to VPI including hypernasality and audible nasal emission.¹⁶ The Goldman Fristoe Test of Articulation 3 was administered at the same visit.¹⁷ The presence or absence of compensatory misarticulation patterns during standardized testing was reported.

Nasopharyngoscopy:

Nasopharyngoscopy exams were conducted using a 2.4mm Pentax Fiber Optic or 2.9mm Distal Chip Nasopharyngolaryngoscope (Pentax Medical, New Jersey) that was inserted with sterile lubricant along the floor of the nose into either the right or left nares. The speech sample obtained varied across patients based on the oral phonemes each patient could accurately produce. Nasopharyngoscopy exams were recorded with synchronous audio and video.

Nasopharyngoscopy recordings were cropped to include only the portion of the examination containing oral speech stimuli. Two trained speech-language pathologists with over 15 years of experience in assessing resonance disorders completed a calibration training session which included viewing and discussing 10 non-study nasopharyngoscopy exams and then independently rating 5 non-study scopes (100% agreement for identification of notching). After the calibration session, raters independently rated the exams indicating if notching was present or absent on the nasal velar surface during rest and/or during phonation. Figure 1 demonstrates examples of patients with and without velar notching. If raters disagreed on the presence or absence of velar notching, a third trained rater with similar experience in assessing resonance disorders independently rated the exam to reach a consensus for analysis. Disagreement was present on six of the 44 exams. Consensus was reached using the third rater in four of these six exams. Consensus could not be reached for the remaining two exams, and these exams were coded as unable to rate.

Figure 1—

Examples of six different patients with three displaying notching in the nasal velar surface (top row; arrow marks the notch) and three showing no notching (bottom row).

Magnetic Resonance Imaging:

Patients completed a whole head MRI on a 3T Phillips Scanner using a fully awake, non-sedated, non-contrast protocol. The imaging protocol included a high-resolution T2-weighted turbo-spin-echo 3D sequence obtained at rest, followed by sagittal and oblique coronal T2 images obtained during sustained /i/ phonation and then sustained /s/ phonation. If the patient could not accurately articulate /s/, images were obtained during phonation of /ʃ/ or at rest only. A single rater with over 15 years of experience in MRI evaluations of velopharyngeal anatomy conducted the measures on MRI data. The rater was blind to nasopharyngoscopy results and was not involved in the direct care of any of the patients. Variables measured included: (1) distance from the LVP sling in the velum to the posterior border of the hard palate and (2) a rating of the continuity of the LVP sling to include (a) completely cohesive, (b) incohesive in the anterior region of the LVP sling (c) incohesive in the middle region of the LVP sling, and (d) incohesive in the posterior LVP sling. Ratings of cohesiveness could include more than one qualifier, such as a case where the LVP sling was incohesive in the middle and posterior region. Figure 2 demonstrates the two measures taken and examples of the levels of cohesiveness of the LVP muscle.

Figure 2.

Demonstration of an MRI midsagittal image showing a schematic representation of the levator veli palatini (LVP) muscle coursing from the base of the skull into the body of the velum. The LVP cohesiveness (top right image) was evaluated as degree of muscle cohesiveness in the (A) anterior, (B) middle (region continuous with the long axis of the LVP muscle bellies), and (C) posterior region of the LVP sling. The distance from the posterior palate to the LVP muscle (bottom right image) was calculated as the distance from the posterior hard palate to the point of the LVP insertion into the body of the velum, along the axis of the muscular sling.

Statistical Analysis:

Fleiss’ kappa was used to measure the reliability of agreement between nasopharyngoscopy raters. Sensitivity, specificity, positive predictive value, and negative predictive value were reported with 95% confidence intervals. Comparisons between patients with and without notching on nasopharyngoscopy were conducted using the Mann-Whitney Test for continuous variables.

Results

A total of 44 patients between the ages of 3–39 years of age met the inclusion criteria for the study. After excluding five exams due to inability to determine presence or absence of notching and removal of the two exams in which consensus was unable to be reached, 37 nasopharyngoscopy exams were included for analysis. Twelve exams (32%) had notching, while 25 (68%) did not. Inter-rater reliability between the two primary raters was k=0.75, indicating substantial agreement.¹⁸

Table 1 summarizes the study population. Although factors such as race, language, and ethnicity likely have no impact on the presence of notching, patient demographics are reported to provide a comprehensive view of the patients enrolled in the study.¹⁹ Patients did not meaningfully differ between notching groups. All patients with notching present had a history of overt cleft palate. Among patient with notching absent, 44% had an overt cleft palate, 20% had a history of overt submucous cleft palate, and 36% had a normal appearing palate on exam (i.e., occult submucous cleft palate or non-cleft VPI). Those with without notching on nasopharyngoscopy were more likely to have moderate hypernasality (68% versus 42%), but less likely to have severe hypernasality (16% versus 33%) than the group with notching.

Table 1.

Patient Demographics

	Notching Present (n = 12)		Notching Absent (n = 25)
	n	%	n	%
Gender
Male	9	75	15	60
Female	3	25	10	40
Race
White or Caucasian	11	92	23	92
Asian	1	8	1	4
American Indian/Alaska Native	-	-	1	4
Ethnicity
Hispanic or Latino	4	33	9	36
Non-Hispanic or Latino	8	67	16	64
Language1
English	12	100	23	92
Spanish	2	17	4	16
Other	-	-	2	8
Cleft Type
Partial signs of SMCP2, 3	-	-	2	8
All clinical signs of SMCP	-	-	3	12
Veau I	1	8	2	8
Veau II	4	33	1	4
Veau III	5	42	5	20
Veau IV	1	8	3	12
Unknown veau type	1	8	-	-
No known cleft palate or SMCP	-	-	9	36
Pierre Robin Sequence	3	25	-	-
Syndrome
22q11.2 deletion syndrome	-	-	11	46
Craniofacial Microsomia	-	-	2	12
Stickler Syndrome	1	8	-	-
Other	1	8	2	8
Age
Median (q1, q3)	12	(10, 18)	9	(6, 16)
Hypernasality Rating
Minimal	1	8	1	4
Mild	2	17	3	12
Moderate	5	42	17	68
Severe	4	33	4	16
Audible Nasal Emission Rating
Absent	4	33	10	40
Occasionally present	3	25	6	24
Frequently present	5	42	9	36
CMAs4 present in speech
Yes	5	42	8	32
No	5	42	16	64
Not reported	2	16	1	4

Notes:

¹Some patients reported more than one language spoken in the home

²submucous cleft palate (SMCP)

³Clinical signs of an unrepaired SMCP include a bifid uvula, zona pellucida, and hard palate notching. A partial SMCP indicates one or two of the clinical signs were present.

⁴Non-oral compensatory misarticulation patterns (CMAs)

Additional footnote: Categorical measures reported as n (%) and continuous measures reported as median (q1, q3).

To determine the accuracy of notching on the nasal velar surface as an indicator for LVP muscle continuity, sensitivity and specificity were calculated with LVP muscle continuity as measured via MRI considered the gold standard for comparison (Table 2). The sensitivity of a velar notch on nasopharyngoscopy for detecting a partially or totally discontinuous LVP on MRI was 43% (95% CI 22–66%). In other words, among those with a partial or total LVP dehiscence on MRI, presence of a notch correctly identified discontinuity in the LVP 43% of the time.

Table 2:

Sensitivity & specificity by location of LVP discontinuity.

			LVP Continuity on MRI
			Discontinuous	Continuous
Complete & Partial LVP Discontinuity	Notching on NP	Present	True positive = 9	False positive = 3
		Absent	False negative = 12	True negative = 13
		Sensitivity	43% (95% CI 22–66%)	NA
		Specificity	NA	81 (95% CI 54–96%)
Anterior LVP Discontinuity Only	Notching on NP	Present	True positive = 1	False positive = 11
		Absent	False negative = 4	True negative = 21
		Sensitivity	20% (95% CI 1–72%)	NA
		Specificity	NA	66% (95% CI 47–81%)
Middle LVP Discontinuity Only	Notching on NP	Present	True positive = 5	False positive = 7
		Absent	False negative = 6	True negative = 19
		Sensitivity	45% (95% CI 17–77%)	NA
		Specificity	NA	73% (95% CI 52–88%)
Posterior LVP Discontinuity Only	Notching on NP	Present	True positive = 8	False positive = 4
		Absent	False negative = 10	True negative = 15
		Sensitivity	44% (95% CI 22–69%)	NA
		Specificity	NA	45% (95% CI 54–94%)

LVP = levator veli palatini; NP = nasopharyngoscopy

The specificity of absent velar notching on nasopharyngoscopy for detecting a completely continuous LVP on MRI was 81% (95% CI 54–96%). In other words, among those with a completely continuous LVP on MRI, the absence of a notch correctly identified continuity 81% of the time. Velar notching as an indicator of LVP discontinuity specific to the anterior, middle, or posterior portion of the LVP was also examined, with similar results (Table 2).

Predictive values provide a reference point to compare the probability of a positive or negative notch finding to indicate LVP status as viewed on MRI. Positive predictive value (PPV) is the probability of a present notch on nasopharyngoscopy correctly identifying a discontinuous LVP on MRI. In the current cohort, the PPV of the presence of notching for identifying a partially or totally discontinuous LVP was 75% (95% CI 49–90%). In contrast, the negative predictive value (NPV) is the probability of an absent notch correctly identifying a continuous LVP. The probability of the absence of notching correctly identifying a completely continuous LVP was 52% (95% CI 41–63).

On MRI, the distance from the posterior palate to the LVP muscle is an additional indicator of insertion status of the LVP, with smaller distance suggestive of anterior insertion of the LVP.²⁰ Median distance from the posterior palate to the LVP muscle in patients with notching noted on nasopharyngoscopy was 9.8 mm (IQR 7.4–12.4). Among patients with absence of notching, median distance from the posterior palate to the LVP muscle was 10.5 mm (IQR 7.0–12.5). No significant difference in distance from the posterior palate to the LVP muscle was present between those with and without a notch (p = 0.987, Mann-Whitney Test).

Discussion

Results from this study call into question the validity of using velar notching on nasopharyngoscopy as an indicator of LVP discontinuity or anterior positioning. Only 43% of patients with a discontinuous LVP had velar notching – for teams that want to offer palatoplasty when the LVP is discontinuous, they will miss over half of eligible patients if they use velar notching as their indicator of LVP discontinuity. Additionally, the presence of notching was not associated with LVP position in the palate, indicating the presence of velar notching does not accurately identify patients with anteriorly inserted LVPs. Taken as a whole, these results suggest cleft teams should use caution when relying on velar notching to direct their VPI treatment plan.

The low sensitivity of velar notching for presence of LVP discontinuity, at 43%, indicates that notching fails to accurately detect LVP discontinuity. This is evidenced by the large proportion of false negatives, patients without notching on nasopharyngoscopy but who demonstrate partial or total LVP dehiscence on MRI. The higher specificity of 81% indicates that when notching is not identified on nasopharyngoscopy there can be more confidence that the LVP is continuous; however, there is still a clinically relevant chance that MRI will show a discontinuous LVP sling even when no notching is present.

Sensitivity in a notch detecting LVP discontinuity was the lowest for the anterior region of the LVP sling. This observation may be due to the nasopharyngoscopy scope being positioned at or posterior to the anterior region of the LVP sling, limiting visualization of the anterior region of the velum. A limitation of nasopharyngoscopy is the difficulty in controlling for lens placement and therefore this theory cannot be confirmed. The position of the LVP muscle relative to the velar knee could also contribute to difficulty identifying anterior LVP discontinuity on nasopharyngoscopy. As described by Boorman and Sommerlad (1985), the velar dimple (noted on the oral side of the velum during velar elevation) is located at the posterior limits of the LVP muscle.²¹ As such, a notch on the anterior region of the LVP sling may be less visible because the anterior region of the LVP is not elevated to the same extent as the middle or posterior regions of the LVP during phonation.

It was expected that velar notching on nasopharyngoscopy would be predictive of LVP muscle continuity on MRI because the LVP muscle courses though the region where the notch is visibly present. An inspection of MRI data did not support this: in 6 out of the 12 patients with velar notching, the velar notch was visible on MRI and the notch was present in the same plane as the LVP muscle sling (Figure 3). Due to the depth distortion and variable camera location for nasopharyngoscopy, it is not possible to determine the location of the notch relative to the LVP sling using nasopharyngoscopy. However, MRI can provide such details.

Figure 3—

Demonstration of the variable presentations of LVP status across the conditions of absent notching and present notching. The presence or absence of notching seen on the scopes are also visible on the MRI, however, the cohesiveness of the LVP is variable across patients, regardless of notching status.

MRI versus nasopharyngoscopy findings in two subgroups of our study population warrant further mention, a patient with unrepaired SMCP and those with no history of any type of cleft palate (noncleft VPI). For the patient with an unrepaired SMCP, a notch was not identified on nasopharyngoscopy, but a discontinuous LVP was observed on MRI). This finding is clinically relevant, as sagittally-oriented LVP muscle fibers have been described as a key feature among those with SMCP and these discontinuities were not associated with a velar notch.⁸ Of the nine patients with noncleft VPI, none of them had a notch observed on nasopharyngoscopy, as would be anticipated. These patients also had completely continuous LVPs on MRI. Taken together, results of patients with a high likelihood of discontinuous LVPs (SMCP) and those with a very low likelihood of discontinuous LVPs (non-cleft) further support our findings of low sensitivity and relatively stronger specificity when comparing LVP morphology between MRI and nasopharyngoscopy.

Because the purpose of this study was not to evaluate speech outcomes, further research is needed to fully understand the clinical relevance of underlying anatomic differences and predictive factors for a successful speech outcome across patient populations. Williams et al. (2022) report anatomic observations from MRI among 113 patients with VPI.²² Common MRI findings among patients with repaired cleft palate (with or without cleft lip) and persistent VPI included short velum, discontinuous LVP sling, and decreased distance from the posterior palate to the LVP muscle. These findings further support LVP morphology as a salient characteristic of residual VPI and a promising intervention target when planning surgical approach. Taken in the context of current study results, it will be important for future research to continue to build a body of evidence that supports optimal imaging for accurate description of LVP morphology and confirms restoration of LVP continuity as positively associated with improvement in velopharyngeal function post-operatively.

A limitation of the present study is the small sample size, particularly the small cohort of patients with velar notching. Future studies should examine these results using a larger sample. In addition, velar notching is poorly defined and as such, raters may use different criteria when determining the presence of velar notching. This may substantially alter the results; however, in the present study we conducted a calibration session to limit the impact of variable definitions among raters and we established definitions a priori. Additionally, nasopharyngoscopy image quality varies based on scope, user, and patient. Our results may not generalize to data from cleft teams obtaining substantially higher or lower image quality than that of the present study. MRI is considered to the be the gold standard for imaging and measuring soft tissue structures, particularly skeletal muscles.^23–26 For this reason, we similarly used MRI as the gold standard for defining variability in the LVP muscle, as opposed to observations at the time of surgery. Future research should consider the agreement between intraoperative and MRI assessments of the LVP status.

Conclusion

Results from this study demonstrate that the observation of a velar notch, as seen on nasopharyngoscopy, is not an accurate method for drawing conclusions about the LVP muscle continuity or relative position within the palate. More specifically, nasal velar notching on nasopharyngoscopy does not consistently identify LVP discontinuities that are visible on MRI. Absence of notching is more accurate in identifying LVP continuity than presence of notching is in identifying LVP discontinuity.

MRI is the only imaging tool that can directly visualize the velopharyngeal muscles in vivo. Caution should be taken when drawing conclusions about the LVP muscle based on imaging methods that are not able to directly visualize the VP muscles. Future research is needed to examine other imaging tools such as videofluoroscopy and to examine whether a combination of imaging modalities and/or findings would increase sensitivity when compared to MRI.