Diagnostic accuracy of lateral cephalograms and cone-beam computed tomography for the assessment of sella turcica bridging

Introduction

The purpose of this research was to assess the diagnostic accuracy of sella turcica bridging on lateral cephalograms when compared with true sella turcica bridging determined via cone-beam computed tomography (CBCT).

Methods

A cross-sectional study was conducted using CBCT images from which lateral cephalograms were generated. The study included 185 subjects (118 females and 67 males; age range, 10-30 years; mean age, 16.63 ± 4.20 years). Sella turcica landmarks and related measurements were calculated for both diagnostic modalities and analyzed by 1 examiner. Subjects were classified into 1 of 3 outcome groups: no bridging, partial bridging, and complete bridging. Diagnostic accuracy was evaluated using sensitivity, specificity, positive and negative predictive values, and receiver operator characteristic curves.

Results

Ten patients were diagnosed as complete bridging on CBCT, whereas 31 patients were diagnosed as complete bridging on lateral cephalogram. Although the lateral cephalogram detected all subjects with complete bridging, it incorrectly classified 12% of subjects. The percent agreement between both diagnostic methods was 55.68%, with a kappa statistic of 0.22 on the right sella turcica and 0.20 on the left sella turcica, indicating fair but statistically significant agreement. The overall accuracy of lateral cephalograms as a diagnostic modality in discriminating between no bridging and partial bridging was good as determined with the area under the curve values of 0.86 and 0.85 for right and left sides, respectively.

Conclusions

Although lateral cephalograms overestimate patients with complete bridging compared to CBCTs, they are a suitable screening modality for accurately suggesting complete sella turcica bridging and differentiating between patients with no bridging and partial bridging.

Highlights

  • The frequency of complete sella bridging was 5.4% on CBCT and 16.8% on lateral cephalogram.

  • Lateral cephalograms are suitable for accurately suggesting complete sella turcica bridging.

  • Lateral cephalogram can accurately discriminate between no bridging and partial bridging.

The development of sella turcica and its morphologic variants has been extensively studied in the literature. , The sella turcica is a saddle-shaped depression on the intracranial surface of the sphenoid bone. There are 5 different morphologic variations of sella turcica as reported by Axelsson et al : oblique anterior wall, sella turcica bridging, double contour of the floor, irregularity in the posterior dorsum sella, and pyramidal shape of dorsum sella.

The sella turcica bridge is a complete ossification of the interclinoid ligaments, which may occur either unilaterally or bilaterally. Although the exact cause of bridging is not known, multiple theories have been proposed, including an abnormality in the embryologic development of the sphenoid bone resulting in bridging, ossification of the dura mater between the anterior and posterior clinoid processes, or because of focal infections of the pituitary gland.

This normal variant of sella turcica occurs with a reported incidence of 1.1%-13% of the general population, with an increased prevalence in those with severe craniofacial disproportions. , In addition, several studies have reported an increased frequency of bridging in adolescents with dental anomalies, such as canine impaction (odds ratio, 3.93; 95% confidence interval [CI], 1.43-10.7) and tooth transposition (odds ratio, 2.9; 95% CI, 1.1-7.9). As calcification in the sella turcica region can appear in early childhood, recognizing sella turcica bridging on diagnostic imaging may allow for early diagnosis and timely intervention of palatal canine impactions, potentially reducing the complexity of treatment. As well, it can serve as a screening method for rare genetic diseases such as Axenfeld-Rieger syndrome with PITX2 mutation. This is an autosomal dominant disorder that is caused by the abnormal migration of neural crest cells. Although the phenotype in this syndrome is heterogenous, midface hypoplasia, dental hypodontia, and sella turcica anomalies are commonly present. ,

Because of the 2-dimensional (2D) view provided in conventional lateral cephalograms, it is difficult to ascertain a clear distinction between a sella turcica bridge when there is complete bony fusion and the appearance of fusion between the anterior and posterior clinoid processes because of radiographic superimposition. In previous autopsy studies, the occurrence of a true sella turcica bridge varied between 2% and 6%. However, sella turcica bridging diagnosed on lateral cephalograms is reported at a higher frequency because of the difficulty distinguishing between a true sella turcica bridge and a pseudobridge. As a result, previous studies have questioned the reliability of using 2D radiographs for accurate diagnosis of sella turcica bridging. ,

Technological advances in 3-dimensional (3D) imaging using cone-beam computed tomography (CBCT) offer noteworthy advantages in the quality and quantity of anatomic data. 3D CBCT has the advantage of overcoming challenges of superimposition and differential magnification of bilateral structures. Furthermore, 3D CBCT scans acquired from a patient can be used to generate 2D lateral cephalograms, thus minimizing further cost and radiation exposure to the patient. , The objective of this cross-sectional study was to assess the diagnostic accuracy of sella turcica bridging on conventional lateral cephalograms when compared with true sella turcica bridging determined via CBCT (considered the reference standard).

Material and methods

This cross-sectional study was approved by the State University of New York at Buffalo Health Sciences Institutional Review Board (no. STUDY00001251). The sample included CBCT images of patients ranging from 10 to 30 years of age assessed between January 2005 and August 2015 at 1 oral and maxillofacial surgery practice.

The inclusion criteria were full volume CBCT images, clear representation of sella turcica, and any type of malocclusion. Records of subjects with a significant pathology in the maxillofacial region, clear craniofacial syndromic anomalies, and CBCT images with motion artifacts were all excluded.

CBCT images were acquired using an i-CAT Cone Beam 3D dental imaging system (version 3.1.62; Imaging Sciences International, Hatfield, Pa) at these settings: 3-7 mA, 120 kV, exposure time of 40 seconds, voxel size of 0.4 mm, and a focal spot of 0.5 mm and a scanning area of 16 × 13 cm. The majority of scans were taken at 0.4 mm voxel size for 40 seconds with a greyscale range of 14 bits, except for 3 scans taken at 0.3 mm voxel size for 40 seconds.

The CBCT images were obtained by an experienced technician. All patients were seated in an upright position with their heads oriented with the occlusal plane parallel to the floor. Head and chin support were used to stabilize the head position, and all teeth were out of occlusion by having the patient bite on a cotton roll. The images were exported in digital imaging and communications in medicine format then imported into Dolphin 3D Imaging System (version 11.7.05.66 Premium; Dolphin Imaging and Management Solutions, Chatsworth, Calif) for analysis.

The original CBCT volumes were standardized by setting the orientation of the axial (x), midsagittal (y), and coronal (z) planes. The axial plane (x) was set to the Frankfort Horizontal plane. The midsagittal plane (y) was defined by nasion (N), anterior nasal spine (ANS), and basion (Ba) landmarks. Then, the multiplanar views were configured as 0.4 mm slices to match the thickness of the original scan.

Synthesized lateral cephalograms were generated from each standardized CBCT volume in the perspective projection, using mechanical porion as the projection center, to simulate the geometry of the conventional lateral cephalograms. Because the original CBCT volumes were oriented to Frankfort Horizontal, synthesized lateral cephalograms were also oriented to Frankfort Horizontal. The Dolphin 1 filter was then used to improve image sharpness.

An onscreen 0.5 mm marker was used to identify each landmark in the multiplanar views and 3D rendered image ( Table I and Fig 1 ). A total of 6 landmarks were identified, and 4 measurements were calculated. Coordinate data for each landmark was used to calculate the Euclidean distance between the 2 paired points. These landmarks were not measured directly on the CBCT reconstruction.

Table I
Sella turcica landmarks and measurements on CBCT and lateral cephalograms
Landmark Definition Measurement Definition
CBCT
TS Tuberculum sellae; midpoint on the anterior boundary of sella identified on the midsagittal plane TS to DS Length of sella turcica
DS Dorsum sellae; midpoint on the posterior boundary of the sella turcica on the midsagittal plane Right ACP-PCP Interclinoid distance on right side
R ACP Apex of the anterior clinoid process on the right side Left ACP-PCP Interclinoid distance on left side
L ACP Apex of the anterior clinoid process on the left side Complete sella turcica bridging Distance between ACP-PCP equals zero
R PCP Apex of the posterior clinoid process on the right side
L PCP Apex of the posterior clinoid process on the right side
ACP Apex of the anterior clinoid process
PCP Apex of the posterior clinoid process
Lateral cephalograms
TS Anterior boundary of sella turcica TS to DS Length of sella turcica
DS Posterior boundary of sella turcica ACP-PCP Interclinoid distance
ACP Apex of the anterior clinoid process
PCP Apex of the posterior clinoid process

R , right; L , left.

Fig 1
Identification of landmarks on CBCT: Multiplanar slices.

To ensure the investigator was blinded, random case identification numbers were generated and reordered for landmark identification on lateral cephalograms, which occurred 2 weeks after landmarks were identified on CBCT images ( Table I ). Distances between the landmarks were calculated using the 2 Pt Line measurement tool in the Dolphin 3D Imaging software. ( Fig 2 ).

Fig 2
Identification of landmarks on lateral cephalogram: A, distance calculated between ACP and PCP; B, distance calculated between TS and DS.

Quantification of sella turcica bridging was performed using the method of Sundareswaran and Nipun, which uses the ratio of interclinoid distance (ACP-PCP) to length (TS-DS) to classify the extent of interclinoid calcification into 3 categories: no bridging (ratio, ≥33%), partial bridging (ratio, >0%-33%), and complete bridging (ratio, 0%). For the CBCT images, sella turcica bridging was quantified for the right and left sides of the sella turcica separately ( Fig 3 ). For the lateral cephalograms, the measurements were conducted without differentiation between sella turcica sides. The data were collected, analyzed, and measured by 1 investigator (A.M.A) who was trained in sella turcica landmark identification on CBCT images by an experienced investigator.

Fig 3
Depiction of different sella turcica bridging types on CBCT: A, complete bridging; B, partial bridging; and C, no bridging.

The interclass correlation coefficient (ICC) was calculated to assess the intraexaminer reliability. Two weeks after from initial examination, 10 randomly selected radiographs were remeasured on both diagnostic modalities. The same 10 radiographs were also compared against an experienced orthodontist to assess interexaminer reliability. The method error was determined using Dahlberg’s formula.

Statistical analysis

Data were analyzed using SPSS Software for Windows (version 23.0; IBM, Armonk, NY). Subjects were classified into 1 of the 3 outcome groups by calculating the ratio between ACP-PCP/TS-DS and using a cutoff of 33%. Diagnostic accuracy of lateral cephalograms was evaluated using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operator characteristic curves for outcome groups. Agreement was assessed using the kappa static.

Results

A total of 218 records were assessed, and 32 were excluded for lack of visualization of sella turcica and one because of the possibility of pathology. The study included 185 subjects (118 female and 67 male) with a mean age of 16.63 ± 4.20 years. There was an approximate difference of 9 months between mean ages for males and females (females, 16.36 ± 4.10 years; male, 17.10 ± 4.59 years).

There was a high agreement for the repeated measurements between the 2-time points on CBCT (ICC: length, 0.972; interclinoid distance R, 0.990; interclinoid distance L, 0.989) and lateral cephalogram (ICC: length, 0.979; interclinoid distance, 0.992). In addition, there was a high agreement between both examiners with all coefficients near the maximum of one (ICC: length, 0.961; interclinoid distance R, 0.996; interclinoid distance L, 0.998). The average error is depicted in Table II .

Table II
Mean error of measurements according to Dahlberg’s formula
Variable Method error
Distance TS-DS 0.26
Interclinoid distance R 0.24
Interclinoid distance L 0.20
Interclinoid distance (ACP-PCP) 0.20
Sella turcica length (TS-DS) 0.23
ACP-PCP/TS-DS 0.02

R , right; L , left.

The average interclinoid distance on lateral cephalogram was 4.01 mm and the sella turcica length (TS-DS) was 10.58 mm ( Table III ). With respect to the measurements taken on CBCT, the average interclinoid distances were 6.14 mm and 6.10 mm for the right and left sides, respectively. The average sella turcica length (TS-DS) was 10.13 mm.

Aug 14, 2021 | Posted by in Orthodontics | Comments Off on Diagnostic accuracy of lateral cephalograms and cone-beam computed tomography for the assessment of sella turcica bridging

VIDEdental - Online dental courses

Get VIDEdental app for watching clinical videos