Computational Analysis of the Mature Unilateral Cleft Lip Nasal Deformity on Nasal Patency.

BACKGROUND: Nasal airway obstruction (NAO) due to nasal anatomic deformities is known to be more common among cleft patients than the general population, yet information is lacking regarding severity and variability of cleft-associated nasal obstruction relative to other conditions causing NAO. This preliminary study compares differences in NAO experienced by unilateral cleft lip nasal deformity (uCLND) subjects with noncleft subjects experiencing NAO.
METHODS: Computational modeling techniques based on patient-specific computed tomography images were used to quantify the nasal airway anatomy and airflow dynamics in 21 subjects: 5 healthy normal subjects; 8 noncleft NAO subjects; and 8 uCLND subjects. Outcomes reported include Nasal Obstruction Symptom Evaluation (NOSE) scores, cross-sectional area, and nasal resistance.
RESULTS: uCLND subjects had significantly larger cross-sectional area differences between the left and right nasal cavities at multiple cross sections compared with normal and NAO subjects. Median and interquartile range (IQR) NOSE scores between NAO and uCLND were 75 (IQR = 22.5) and 67.5 (IQR = 30), respectively. Airflow partition difference between both cavities were: median = 9.4%, IQR = 10.9% (normal); median = 31.9%, IQR = 25.0% (NAO); and median = 29.9%, IQR = 44.1% (uCLND). Median nasal resistance difference between left and right nasal cavities were 0.01 pa.s/ml (IQR = 0.03 pa.s/ml) for normal, 0.09 pa.s/ml (IQR = 0.16 pa.s/ml) for NAO and 0.08 pa.s/ml (IQR = 0.25 pa.s/ml) for uCLND subjects.
CONCLUSIONS: uCLND subjects demonstrated significant asymmetry between both sides of the nasal cavity. Furthermore, there exists substantial disproportionality in flow partition difference and resistance difference between cleft and noncleft sides among uCLND subjects, suggesting that both sides may be dysfunctional.

INTRODUCTION

The anatomic deformity associated with unilateral cleft lip nasal deformity (uCLND) creates multiple sites of nasal airway obstruction (NAO), such as septal deviation, nostril atresia, valvular stenosis or deficiency, turbinate hypertrophy, and altered nasal floor from deficiency of maxillary growth.[1-6] These deformities impair nasal breathing, which negatively impact patients’ quality of life.[1,7-11] Patients with uCLND routinely suffer from sleep-disordered breathing, they snore frequently and experience loud labored breathing during sleep.[12-21] Furthermore, some patients report air hunger and trouble breathing during exercise.[22]

The functional implications of uCLND have been poorly understood and underappreciated.[22] Nonetheless, majority of cleft lip and/or cleft palate studies that have addressed NAO have focused on a comparison with normal subjects and/or among different cleft phenotypes.[6-10,22-29] Although these studies have demonstrated the impact of uCLND on nasal breathing difficulty, they do not provide insights into variabilities in nasal dysfunction across cleft and noncleft subjects with NAO. The cleft literature is deficient on information comparing the extent of impaired nasal breathing in cleft-associated nasal obstruction with other conditions associated with nasal obstruction. Quantifiable evidence pertaining to differences in nasal breathing patterns across different disease conditions is important because it will provide the cleft and craniofacial community with objective evidence regarding extent of the effect of uCLND on nasal breathing and provide perspective on the importance of treatment.

Innovative assessment tools to objectively describe NAO may provide enhanced detail for understanding uCLND-associated nasal obstruction. Computational fluid dynamics (CFD) methods have shown significant promise in describing upper airway physiology.[30-35] They allow for the merger of anatomy and physiology by creating anatomically realistic 3-dimensional (3D) nasal cavity models from patient-specific radiographs with computed measures of airflow, heat transfer, and air humidification.[35-41] The purpose of this preliminary study is to use CFD modeling to quantify nasal patency differences associated with nasal obstruction in uCLND patients compared with noncleft patients who had a clinical diagnosis of nonreversible, surgically treatable NAO.

MATERIAL AND METHODS

Selection of Study Cohort

This is a retrospective study approved by the Duke University Health System Institutional Review Board for Clinical Investigations. Twenty-one subjects with high resolution computed tomography (CT) scans were selected based on a chart review of medical records from 2010 to 2017. Selected subjects were classified into 3 groups:

(1) Five subjects (25–70 years, 2 males, and 3 females) with radiographically healthy normal nasal cavity and sinus anatomy (NORMAL). These subjects had no nasal airway functional symptoms.

(2) Eight noncleft subjects (20–55 years, 5 males, and 3 females) with a clinical diagnosis of nonreversible, surgically treatable cause for NAO such as septal deviation, turbinate hypertrophy resistant to medical therapy, and lateral nasal wall collapse (NAO; Table 1).

Table 1:

Basic Demographic Information and Diagnoses

(3) Eight subjects (15–35 years, 2 males, and 6 females) with a clinical diagnosis of uCLND and have not had definitive cleft rhinoplasty or septoplasty for correction of NAO (CLEFT; Table 1).

Nasal Airway Reconstruction

Subject-specific and anatomically realistic 3D nasal cavity models were created from radiographic images of CT scans for all subjects. Nasal cavity models were segmented, from CT to create 3D nasal models, similar to previously published models by our group.[33,34,37,39,40,42-44] Next, 3D nasal cavity models were imported into the mesh-generating software, ICEM-CFD 16.1 (ANSYS, Inc., Canonsburg, Pa.) for creation of planar inlet surface at the nostrils, a 2-cm tube attached to the nasopharynx, and an outlet surface at the lower end of the tube. To solve the discretized governing equations of fluid flow, each nasal cavity model was discretized by generating approximately 4 million tetrahedral mesh elements inside the airway based on a mesh density refinement analysis study in Frank-Ito et al.[40]

Nasal Airflow Simulations

Following discretization of nasal models, the CFD software package Fluent 16.1 (ANSYS, Inc., Canonsburg, Pa.) was used to conduct steady-state simulations of laminar incompressible inspiratory airflow at resting inhalation rates. ANSYS Fluent uses the finite volume method to numerically solve the discretized Navier–Stokes equations. Because the flow velocities through the nasal passage during low-to-moderate breathing rate usually have a Mach number ,[45] the conservation of mass and momentum governing equations for laminar, incompressible steady-state flow reduces to

where is the velocity vector, kg/m3 is fluid density, is dynamic viscosity, and is pressure. Airflow was simulated in each nasal cavity using these boundary conditions: at the nasal wall, no-slip, and stationary wall; at the inlet, a “pressure-inlet” boundary condition was specified at the inlet surface with gauge pressure set to zero; and a “pressure-outlet” condition at the outlet with a negative gauge pressure set to target resting breathing of 15 l/min flow rate.

Computed Outcomes and Analysis

Analyses of nasal cavity anatomy (shape and size) and airway function (airflow profile) were conducted across the 3 categories of subjects (NORMAL, NAO, and CLEFT). Data computed for outcomes of nasal cavity shape and size included: (1) cross-sectional images of the airway at 11 cross sections from immediately after the posterior end of nostrils tip to choana (Fig. 1); (2) unilateral left–right absolute difference in cross-sectional area at every cross section; and (3) unilateral shape factor circularity at each cross section of the nasal cavity. Shape factor circularity tells the degree to which each unilateral cross section is similar to a circle; it is a measure that describes the form and roughness/irregularity of a shape. Circularity is dimensionless and it is defined as where is unilateral cross-sectional area of airway at each cross section and is unilateral cross-sectional perimeter of airway at each cross section. The circularity of a given shape lies between 0 and 1, shapes with smaller circularity values are less round (and/or more irregular) and higher circularity value shapes are more round (and/or less irregular). The circularity of a circle is 1.

Fig. 1.

Cross-sectional images of the nasal cavity at 11 cross sections from posterior end of nostrils tip to choana for NORMAL, NAO, and CLEFT subjects. INV, internal nasal value.[61]

Data computed for functional outcomes related to nasal airflow and airway patency included: (1) unilateral airflow partition between both (left and right) sides of the nasal cavity; (2) unilateral nasal resistance; and (3) flow streamline visualization of air in the nasal cavity. Unilateral airflow partition (represented in percent) is defined as unilateral flow rate into a particular side divided by total (bilateral) flow rate. Nasal resistance was calculated as (Pa.s/mL) where is the unilateral pressure drop from nostril to choana, and is the unilateral volumetric flow rate. Nasal resistance calculated from nostrils to choana has been reported to capture patients’ symptoms of nasal obstruction more accurately than that calculated from nostrils to posterior portion of the nasopharynx.[46] Furthermore, patient-reported Nasal Obstruction Symptom Evaluation (NOSE) scores for NAO and CLEFT subjects were collected. NOSE is a validated 5-item scale (0: good to 4: bad) with items relating to nasal congestion, blockage, difficulty breathing, trouble sleeping, and air hunger sensation during exercise.

To assess measures of central tendency and dispersion, descriptive statistics such as median and interquartile range (IQR) were reported for computed outcomes due to the relative small patient sample in this study. Furthermore, boxplots and error bar (mean SD) charts were constructed for computed outcomes. Analyses were performed using MATLAB R2016a software (MathWorks, Inc, Natick, Mass.).

RESULTS

Nasal Cavity Anatomy

Cross-sectional nasal cavity images of the anatomy at 11 cross sections depicted distinct morphological differences for NORMAL, NAO, and CLEFT subjects (Fig. 1). These anatomical differences in nasal airway were due to CLEFT- and NAO-related pathological deformities compared with those of NORMAL. Furthermore, there was a significant middle-to-posterior septal deviation in the cleft subject from cross section 4 to cross section 11 (choana; Fig. 1).

Figure 2 shows the results of unilateral cross-sectional area differences describing the degree of asymmetry between the left and right sides of the nasal cavity. Mean (SD) cross-sectional area differences between both unilateral sides across all 11 cross sections ranged from 6.68 6.23 mm2 (at cross section 2) to 28.95 21.483 mm2 (at cross section 5) for NORMAL subjects. The range for NAO subjects was 26.93 24.61 mm2 (at cross section 10) to 54.95 33.12 mm2 (at cross section 2) and 29.18 19.69 mm2 (at cross section 2) to 75.96 54.48 mm2 (at cross section 11) for CLEFT subjects. The degree of asymmetry was the greatest for NAO subjects at the anterior region, and was the greatest at the middle-to-posterior region for CLEFT subjects (Fig. 2). In general, CLEFT subjects consistently demonstrated the highest asymmetry for majority of the cross sections.

Fig. 2.

Absolute unilateral left–right cross-sectional area differences at each cross section for NORMAL, NAO, and CLEFT subjects. Error bar is mean ± SD.

As demonstrated in Figure 3A, unilateral circularity for NORMAL subjects ranged from 0.06 0.01 (right-side cross section 7) to 0.70 0.27 (left-side cross section 11). With the exception of cross sections 8, 9, and 10, unilateral circularity values between the left and right nasal cavity were mostly homogenous. Among NAO subjects (Fig. 3B), unilateral circularity ranged from 0.05 ± 0.01 (cross section 7) to 0.52 ± 0.30 (cross section 11) on the less affected side, and from 0.05 ± 0.01 (cross section 5) to 0.55 ± 0.32 (cross section 11) on the more affected side. Unilateral circularity between both sides of the nasal cavity in NAO subjects was very dissimilar at the anterior cross sections (cross sections 1–4) and mostly similar from cross sections 6 to 11. Unilateral circularity among CLEFT subjects ranged from 0.07 ± 0.02 (cross section 6) to 0.28 ± 0.09 (cross section 1) on the noncleft side, and from 0.08 ± 0.02 (cross section 6) to 0.28 ± 0.17 (cross section 1) on the cleft side (Fig. 3C). CLEFT subjects exhibited the greatest dissimilarity in unilateral circularity values between both sides of their nasal cavity; nearly every cross section between the cleft and noncleft sides had sizeable variability.

Fig. 3.

Unilateral shape factor circularity at each cross section of the nasal cavity for (A) Normal Subjects, (B) NAO Subjects, and (C) CLEFT Subjects.

Nasal Airflow Function

The distribution of airflow streamlines are illustrated in Figure 4. Airflow streamlines in the CLEFT nasal cavity traveled via limited and narrowed pathways due to severe nasal occlusion compared to NORMAL (Fig. 4A), and NAO (Figure 4B). (Plots of airflow streamline patterns in the nasal cavity were based on 50 equally spaced and randomly selected seed points on the nostril surface.)

Fig. 4.

CFD simulated nasal airflow streamline pattern in (A) NORMAL subject; (B) NAO subject; and (C) CLEFT subject.

As depicted in Figure 5A, patient-reported NOSE scores were higher in NAO subjects (median NOSE = 75) compared with CLEFT (median NOSE = 67.5) subjects. However, variability in NOSE scores among CLEFT subjects was 7.5 points higher than in NAO subjects; IQR = 30 for CLEFT versus IQR = 22.5 for NAO. Next, airflow partition results in Figure 5B showed that the predominately affected side (AS) and the less affected side (NS) were considerably different among NAO subjects; median values were AS = 34.1% versus NS = 65.9%, with a distribution spread of IQR = 12.5%. Among CLEFT subjects, median values were AS = 39.3% versus NS = 60.7%, with a distribution spread of IQR = 26.8%. (Note that the AS for NAO subjects is the more obstructed size, which is also side where the nasal septum is deviated, whereas AS for CLEFT subjects is the side with unilateral cleft lip deformity.) Furthermore, a pairwise comparison of unilateral airflow partition difference between the left and right nasal cavities in Figure 5C indicated that normal subjects had a median difference of 9.4% (IQR = 10.9%), 31.9% (IQR = 25.0%) for NAO subjects, and CLEFT subjects had a median difference of 29.9% (IQR = 44.1%).

Fig. 5.

A, Boxplot comparing NOSE scores between NAO and CLEFT subjects. B, Unilateral airflow partition for both sides of the nasal cavity for NAO and CLEFT subjects. Among CLEFT subjects, AS is the side with unilateral cleft lip deformity and NS is the noncleft side. C, Unilateral airflow partition difference between both sides of the nasal cavity for NORMAL, NAO, and CLEFT subjects. AS, predominately affected side for NAO subjects; NS, less obstructed size.

Results from Figure 6A revealed that the distribution of unilateral left-to-right difference in nasal resistance was both narrowest and lowest in NORMAL subjects (median = 0.01 pa.s/ml, IQR = 0.03 pa.s/ml), compared with NAO (median = 0.09 pa.s/ml, IQR = 0.16 pa.s/ml) and CLEFT (median = 0.08 pa.s/ml, IQR = 0.25 pa.s/ml), with CLEFT subjects having the largest variability. In addition, as expected, nasal resistance on the AS was higher than on the less affected side (NS) for both NAO and CLEFT subjects (Fig. 6B). However, median nasal resistance difference between AS (0.28 pa.s/ml) and NS (0.17 pa.s/ml) in NAO subjects was 0.11 pa.s/ml; which is higher than 0.05 pa.s/ml, the median nasal resistance difference between AS (0.17 pa.s/ml) and NS (0.12 pa.s/ml) for CLEFT subjects. Suggesting that unlike in NAO subjects, with an obvious predominantly obstructed side, CLEFT subjects do not exhibit such predominantly obstructed side between AS and NS sides (Fig. 6B).

Fig. 6.

A, Unilateral nasal resistance difference between both sides of the nasal cavity for NORMAL, NAO, and CLEFT subjects. B, Unilateral nasal resistance for both sides of the nasal cavity for NAO and CLEFT subjects. Among CLEFT subjects, AS is the side with unilateral cleft lip deformity and NS is the noncleft side. AS, predominately affected side for NAO subjects; NS, less obstructed size.

DISCUSSION

The results of our nasal cavity anatomical comparisons revealed regional cross-sectional airway asymmetry where septal deviation was considerably prominent among the cohort of NAO and CLEFT subjects studied (Fig. 2). The location of deviated nasal septum among NAO subjects was mostly around the anterior to middle portion, whereas CLEFT subjects exhibited greater middle-to-posterior septal deviation in addition to anterior deviation. This finding among cleft subjects is in agreement with reports by Friel et al[25] suggesting that there is a significant degree of deviation at the middle-to-posterior portion of the septum. Nonetheless, surgeries for skeletally mature cleft lip nasal deformity often do not attempt to correct middle-to-posterior septal deviations, corresponding to the osseous septum.[8,9,25] This unrepaired middle-to-posterior deviation accounts for about 75% of the maximal degree of septal deviation within the nasal cavity.[25] Unlike anterior cartilaginous septoplasty, middle-to-posterior osseous septoplasty is relatively more technically challenging to perform and has attendant risks associated with osseous resection, such as mucosal tears, postsurgery bleeding, and cerebrospinal fluid leak.[8,25] However, there is significant deviation around the middle-to-posterior septum that may create residual obstruction after surgery.[25]

As described in Figure 5A, variability in patient-reported NOSE assessment among CLEFT subjects was higher than in NAO subjects. This high variability may allude to the fact that these subjects may be having difficulty self-reporting nasal obstruction symptoms because they have lived without knowing what it is to have normal nasal function.[9,47] Furthermore, for NAO subjects, unilateral flow partition (Fig. 5B) and unilateral resistance (Fig. 6B) between AS and NS were very different. This is in contrast with CLEFT subjects showing a reduced difference between AS and NS, and a high measure of variability based on IQR. This reduced difference in unilateral flow partition and unilateral resistance among CLEFT subjects and high IQR may point to the fact that obstruction on AS may not be dominating the obstruction on NS for CLEFT subjects as was observed in NAO subjects. Rather, CLEFT subjects demonstrated considerable nasal dysfunction on both sides. This finding is consistent with what Shih and Sykes[48] noted, in uCLND the nasal septum is deviated on both sides of the nasal cavity; the septum is deviated to the noncleft side at the anterior portion, and deviated to the cleft side posteriorly.[23,24] In addition, Farzal et al[27] reported no significant difference in unilateral nasal volume between cleft and noncleft sides based on 3D volume quantifications. However, there is disagreement in the literature on the side with greater obstruction as some articles reported differences in unilateral nasal cavity volumes, unilateral cross-sectional areas, and/or unilateral airflow (with the cleft side mostly affected), as measured by radiographic images, acoustic rhinometry, and rhinomanometry.[11,28,29,49,50]

Additionally, Sandham and Solow[49] suggested no significant difference in bilateral nasal resistance between cleft subjects (cleft lip, cleft palate, and unilateral cleft lip and palate) and their control group. Thus, in contrast with many other reports in the literature that cleft subjects experience significant nasal obstruction.[6,8-11,22-29,48,50] Drake et al[26] suggested that the cleft nasal passage was about 30% smaller than noncleft nasal passage. Sobol et al[22] reported that children with cleft lip and cleft palate experience nasal obstructive symptoms at a significantly higher rate than unaffected children. Among the different phenotypes of cleft deformities studied by Sobol et al,[22] children with unilateral cleft lip with cleft palate were significantly affected by symptoms of nasal obstruction; those with bilateral cleft lip with palate and those with cleft palate only were not similarly affected. Peroz et al[47] reported correlations between the volume and size of cleft patients’ nasal airway and their desire for correction of their nasal function; however, these patients did not report increased severity of nasal obstruction.

In most clinical settings, current standards for evaluating cleft-induced NAO include the patient-reported NOSE questionnaires,[51,52] the facial aesthetic patient-reported outcome instrument (FACE-Q)[53] and the Cleft Evaluation Profile,[54,55] and physical examination maneuvers.[9] These subjective measures are highly dependent on patients’ (and/or their primary caregiver) opinions and surgeons’ expert opinion. They do not objectively provide physiologic information such as nasal airflow distribution and nasal resistance. These results demonstrate the strong potential of CFD modeling in quantifying nasal patency and our group has been spearheading this effort. We have used CFD modeling to investigate both (1) patient outcomes after NAO in noncleft patients[36,43,56] and (2) the effects of postsurgery maxillary antrostomy size on sinus aeration in patients treated for chronic rhinosinusitis.[32] In another study, using virtually modified nasal cavity models based on a surgeon’s edits of presurgery scans to mimic actual surgery done, we evaluated whether such models can accurately predict patients’ postoperative nasal outcomes.[41] In Rhee et al,[57,58] we used CFD-based virtual surgery methods to compare computed airflow outcomes from virtual procedures of inferior turbinectomy, septoplasty, and septoplasty with inferior turbinectomy to actual postoperative computed outcomes and to quantify the effects of the following individual virtual procedures on nasal airway patency: nasal valve repair, bilateral inferior turbinectomy, and septoplasty. These findings have provided important information regarding changes in airflow dynamics in the setting of pathology;[32,35,36,43,56] effects of anatomy and nasal disease on topical drug delivery;[33,37,39,40,42,59] and surgical alteration for comparing the contribution of different procedures to nasal function.[34,41,44,60]

CONCLUSIONS

In conclusion, the present study uses computational modeling to provide preliminary objective assessments of the extent of nasal deformity and dysfunction in cleft individuals compared to NAO and normal subjects. Patients with uCLND had significant asymmetry between both sides of the nasal cavity, particularly at the middle-to-posterior region, which is attributable to middle-to-posterior septal deviation. Furthermore, CFD results demonstrate disproportionality in flow partition difference and resistance difference between the cleft and noncleft sides for uCLND patients suggesting that both sides are dysfunctional. Such was not the case for NAO where significantly greater unilateral airflow resistance was reported on the AS (side with septal deviation). Patients with unilateral clefts presenting for treatment of the associated nasal deformity require functional assessment and likely will merit comprehensive functional treatment.

ACKNOWLEDGMENTS

We give special thanks to ANSYS, Drs. Paolo Maccarini, and Murali Kadiramangalam (ANSYS Global Academic Program Director) for support and strategic donation. All authors gave final approval for publication.