INTRODUCTION

Various diagnostic tools such as questioners, endoscopy, the Mueller Maneuver, static radiologic imaging, dynamic scanning, and polysomonography are used for airway evaluation. Polysomonography is deemed as the gold standard test for diagnosing obstructive sleep apnea (OSA).¹ However, it does not provide clinicians with anatomical information to guide therapy. Endoscopy may be used clinically to evaluate the site of airway collapse, but it is not able to evaluate the submucosal and deeper airway morphology and is subjective. Imaging modalities are adjunctive to physical examination and craniofacial characteristics and have been correlated with the severity of airway obstruction. The lateral cephalogram (LC) is considered a standard tool for maxillofacial surgeons to predict treatment interventions and quantify the changes of airway structure before and after the use of mandibular advancement devices or maxillomandibular advancement surgery treatments.^1,2

LC generates two-dimensional (2D) images of a three-dimensional (3D) structure and suffers from distortion, magnification, lack of transverse dimensions, and superimposition of bilateral craniofacial structures,^3,4 which render an inadequate size and a complexity of airway structure assessments. However, LC measurements provide the ability to differentiate between OSA and non-OSA patients and have been correlated with the severity of the disorder. Increasing in upper airway length, narrowing of the nasopharynx and oropharynx, and enlargement of the soft tissue in the upper airway can be correlated with the presence and severity of OSA.^5,6

Computed tomography (CT) provides the ability to evaluate the hard and soft structures of the airway in 3D.⁷ Anatomic measurements such as the retropalatal space and the lateral anteroposterior diameter are correlated with the severity of OSA.^8,9 CT scanning notably enhances soft tissue contrast along with 3D assessment, which allows precise measurements at different airway levels. In addition, CT dynamic scans allow the study of the airway structure changes during respiration and sleep cycling, which can play important roles in evaluating and managing OSA patients. However, high radiation dose exposure (especially for younger people) and high financial costs are considered the main CT scan limitations.^1,10,11

Cone beam computed tomography (CBCT) successfully represents the true 3D morphology of the head and neck structures in the upright position. A CBCT's ability to discriminate borders of soft tissue structure and void spaces makes it a valid and reliable diagnostic modality to analyze static airway structure.¹² Ogawa et al.¹³ have shown a relationship between the airway volume and minimum cross-sectional area in patients with and without OSA. OSA patients presented with less anterior posterior airway dimension, less minimum cross-sectional area, and lower airway volume.

We hypothesize that because the upper airway has a spherical or elliptical shape,^13,14 the linear measurements from 2D images may predict the cross-sectional area and volume of the airway with a significantly smaller dose of radiation exposure than 3D imaging modalities. However, to the best of our knowledge, no reported systematic review is available on the correlation between CBCT/CT and LC airway assessment.

The aim of the present study is to systematically review the literature correlating upper airway measurements from LCs to measurements from CBCT/CT scans. This will help determine the possible utility of LCs as a valid screening tool to estimate airway size to predict OSA risk and quantify the response to surgical intervention.

MATERIALS AND METHODS

Review Protocol

A comprehensive electronic search through PubMed, Cumulative Index to Nursing and Allied Health Literature (CINHAL), Web of Science, and Science Direct databases up to November 8, 2015, was conducted. Table 1 summarizes the final search keywords for each database in detail. A manual search of the reference lists of the included studies was performed to supplement the literature search. Furthermore, to improve the reporting of the current systematic review, Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement guidelines were followed.

Table 1. Key Words Based on Databases in Details

View Fullscreen

Quality Assessment

van Vlijmen¹⁶ performed a systematic of review in 2012 to evaluate what level of evidence is available to reinforce the use of CBCT in orthodontic diagnosis and treatment planning. They developed a scoring system based on Lagravere and Gordons's scoring systems.¹⁷ The scoring system of van Vlijmen's study was adapted for all study designs that evaluated 12 criteria. We adapted the scoring system by removing the randomization item because the studies did not apply a real intervention and are scored using the remaining criteria (Table 2).

Table 2. Methodological Scoring

^a ✓, Fulfills satisfactorily the methodological criteria; o, does not fulfill the methodological criteria or is not clarified; NA, not applicable.

View Fullscreen

The results and mean quality (mQ) were reported and interpreted by percentage as follows: mQ < 60% = poor quality, 60% ≤ mQ < 70% = moderate quality, mQ ≥ 70% = good quality. Finally, any disagreement was settled by discussion and consensus.

RESULTS

Articles

A total of 619 articles were found in the initial review. No study was included through manual search. Of these articles, 18 were considered potentially eligible after title and abstract screening. Upon consultation of the full texts, consequently, 11^18–28 of 18 studies met all the inclusion criteria (Figure 1). Table 3 presents the demographic characteristics of the included studies. Various borders have been applied among studies to define the airway segments. Therefore, in the present review, the segments were considered to be on the topic of nasopharyngeal, oropharyngeal, velopharynx, and hypopharyngeal spaces as determined by the authors.

View Figure

• Download as Powerpoint
• Download Figure

Table 3. Demographic Data of Included Studies^a

^a CBCT indicates cone-beam computed tomography; CT, computed tomography; LC, lateral cephalometry; NS, nonsignificant difference; NR, not reported; VS, voxel size; Con, conventional; FOV, field of view; LR, lateral reconstruction; OSA, obstructive sleep apnea.

View Fullscreen

CT scans – LC correlation

Of the 11 studies, 4 investigated the correlation between CT (supine)/LCs.^18–21 All measurements were reported as none, weak, or moderate correlations, .70 ≤ r ≤ .70 (Table 4).

Table 4. Summary of Results for CT or LC Correlations

^a CT indicates computed tomography; LC, lateral cephalometry; CC, correlation coefficient; naso, nasopharynx; oro, oropharynx; hypo, hypopharyngeal.

View Fullscreen

CBCT scans – LC correlation

Of the 11 studies, 7 evaluated the correlation between the LCs and CBCT (upright) scans in the different segments (Table 5).^22–28

Table 5. Summary of Results for CBCT and LC Correlations

^a CBCT indicates cone-beam computed tomography; LC, lateral cephalometry; CC, correlation coefficient; naso, nasopharynx; oro, oropharynx; hypo, hypopharyngeal, LR, lateral reconstruction of the CBCT images; NS, not significant; ANR, adenoidal nasopharyngeal ration; NP, nasopharynx.

View Fullscreen

Nasopharyngeal segment. A weak to strong correlation with a wide range of correlation coefficients was reported between LCs and CBCTs.^23–28 The adenoidal–nasaopharyngeal ratio in conventional LCs has a negative strong correlation with nasopharyngeal volume in CBCT (r = −.78).²³ Similarly, the linear measurement of PNS to the posterior pharyngeal wall in conventional LCs can predict the nasopharyngeal area (r = .81).²⁴ Moreover, ad2-PNS (upper sagittal depth of the nasopharyngeal airway) in CBCT orthogonal-generated cephalograms can strongly predict the nasopharyngeal volume (r = .93).²⁶

Oropharyngeal segment. A strong correlation (r = .71) was reported between the linear measurements in CBCT-generated cephalograms and area measurements in CBCT scans.²⁶ Also, a strong correlation (r = .83) exists between the area measurement of conventional LCs and the volume measurements of CBCT scans.²² Figure 2 illustrates the strongly correlated measurements.

View Figure

• Download as Powerpoint
• Download Figure

Hypopharyngeal segment. Weak correlations were observed between LC and CBCT.²⁵

Quality Assessment

Of the studies, 4 were of weak quality,^18,19,24,27 5 were of moderate quality,^{21–23,25,28} and 2 were rated to be of strong quality.^20,26 Most of the included studies suffered from a lack of sample size estimation before data collection, selection bias, and lack of reporting confidence intervals (Table 6).

Table 6. Methodological Quality Scores for Study Measurements

^a ✓: Fulfills satisfactorily the methodological criteria. o: Does not fulfill the methodological criteria or is not clarified.

View Fullscreen

DISCUSSION

The present review reveals that there is no strong correlation between LCs in an upright position and CT scans in a supine position. The results may be a result of the dimensional changes in airway that occurs between the supine and upright positions of the patients.^29–31 CT and CBCT scans have been introduced to study the 3D airway structure. However, the effective doses of CT (429.7 μSv) and CBCT (56.2–61.1 μSv) imaging are markedly greater than LCs (10.4 μSv),^32,33 and their long-term effects have still remained unknown.

LC provides static imaging of a dynamic structure, whereas ultrafast CT scanning allows a dynamic evaluation of the airway during respiratory cycles. Of the included studies, Yucel et al. compared dynamic CT imaging with static LC radiographs, which may have contributed to the poor correlation between the modalities.²⁰

In nasopharyngeal segments, no statistically significant difference was found between CBCT scans and conventional LCs in linear measurements.²⁴ Some studies reported a strong correlation between the CBCT and LCs.^23,24,28 However, Aboudara et al.²⁸ used the axial CBCT reconstruction plane passing through PNS as one of the boundaries for the nasopharyngeal area in LCs, which may overestimate the actual efficiency of LCs.

In orthogonally generate LCs from CBCT scans, a strong correlation was reported between linear (LC) and volume (CBCT) measurement, which is in disagreement with the Sears et al.²⁵ study, which reported a weak correlation between CBCT and conventional LCs. The discrepancy may have arisen because orthogonal CBCT projections are more accurate than conventional LC in the midsagittal dimensions.³⁴ However, using the same exposure to generate both images, eliminating the variation in airway size because of function, may explain the higher correlation. In other words, the difference between the LC and the CBCT observed in the Sears et al.²⁵ study might be a result of the difference in tongue position or swallowing rather than a difference resulting from the imaging technique. However, despite the high correlation found between CBCT imaging and LCs in nasopharyngeal segment assessment, LCs are not always accurate. These 2D radiographic images have limitations in displaying the posterior nasopharyngeal airway,³⁵ wide airways, and turbinate protuberance.²⁸ Therefore, using a 3D analysis might be beneficial for such assessments.

The oropharynx is deemed as the most common site of upper airway collapse in adults with OSA.³⁶ Thus, anatomic evaluation of this region is of critical importance. Lenza et al.²⁶ and Bronoosh et al.²² reported a strong correlation between the CBCT scans and LCs. Yet, no strong correlation was found within the other included studies between the CBCT and LC measurements (r ≤ .52) in the velopharyngeal segments.²⁶ The weak correlation in these regions can be explained by the fact that the airway is not spherical except for the nasopharyngeal segment. Moreover, CBCT scans show moderate level of interobserver reliability in linear width, cross-sectional area at the level of the vallecula, and minimum axial area (r = .66).³⁷ Landmarks with lower densities have also a high measurement error in 2D radiographs and 3D imaging.^38,39 However, CBCT images provide more precise soft tissue and hard tissue landmark identification than LCs.⁴⁰

In the hypopharyngeal area, very limited studies have been performed. However, poor correlation exists in this region between the 2D and 3D modalities.^21,25 Poor landmark identification and high superimposition in 2D radiographs can again contribute to the weak correlation between 2D and 3D imaging in this region. In addition, no strong association reported between the total volume of airway obtained through CBCT and LC linear/area measurements. This poor correlation is contributed to dramatic differences in airway shape that exist in the oropharyngeal and hypopharyngeal spaces.¹⁴

The present study reveals the higher correlations for upright CBCT and upright LC than supine CT and upright LC imaging, which may be explained by positional changes in the airway. Reductions in airway antero-posterior dimensions and cross-sectional area have been demonstrated in the supine posture. Furthermore, increasing in the thickness of the soft palate and tongue in the cross-sectional area and increasing in the posterior tongue pressure have been shown in a supine airway position.^29,30,41 In comparison between CT scans in supine position and CBCT scans in sitting upright position, the soft palate, epiglottis, and entrance of the esophagus have a caudal movement when the posture changed from supine to sitting upright and a posterior movement from an upright to a supine position. Hyoid bone caudal movement as a result of postural changes has also been demonstrated.³¹

Lack of control over the confounding factors may justify the inconsistent results. The CBCT scans and LCs ideally should be taken at the same time for accurate comparison, and some included studies did not take this into account.^18,19,25,28 Using the same head positions, tongue positions, techniques, and imaging devices are also essential to reach conclusive results.^42,43 Furthermore, imaging at the different stages of respiration and swallowing cycles between CBCT/CT and LCs may also affect the airway size. During expiration, the airway size at all anatomic segments was significantly larger than inspiration.⁴⁴ In addition, another limitation in the data synthesis of the present review is the various borders that have been defined across the studies. Various anatomic landmarks have been used to border the nasopharyngeal, oropharyngeal, and hypopharyngeal segments, which may justify why some values do not correlate between the studies. However, very close anatomic landmarks have been applied across the most of studies, which may minimize the influence of this confounding factor. (Please contact the authors directly to receive the applied borders in depth within each study.)

Based on the evaluation of the analytical approach section and the presentation of results, it appears that the included studies used Pearson correlation coefficients. However, only a few presented the actual scatterplots of data distribution. Although the appropriate statistical tests seem to have been used, the influence of outliers is unclear. We are then unable to accurately assess if the underlying assumptions (especially data distribution) regarding the use of parametric tests (in this case, Pearson correlation coefficients) were adequately satisfied.

CONCLUSION

• No LC measurements were found to strongly correlate with supine 3D airway measurements obtained thorough CT scans.

• The adenoid nasopharyngeal ratio and the linear measurement from PNS to the posterior pharyngeal wall parallel to floor on conventional LC can be used as the initial screening measurements to respectively estimate the volume and area of the nasopharyngeal upright measurements obtained from CBCT scans.

• The oropharyngeal area measurement from conventional LC can be used as an initial screening measurement to predict the upright upper airway 3D volume.

• Linear measurements of the hypopharyngeal segment from LCs cannot explain most of the variability in the volumetric measurements from CBCT scans.

• CBCT volumetric airway measurements correlated better with airway measurements from CBCT-generated LC than conventional LC. This is likely a result of the fact that the CBCT-generated LC originates from the same exposure as the 3D CBCT image with the muscles surrounding the airway in the same functional position.

• The results of the included studies in this systematic review should be interpreted with caution because the observer reliability, regression analysis scatterplot, and impact of outliers on the correlation were not reported within some of the included studies.