Cone Beam CT Analysis of Haller Cells: Prevalence and Relationship with Orbital Floor Dehiscence

Materials and Methods CBCT images of 120 patients were interpreted in coronal plane for the presence of Haller cells and orbital floor dehiscence. The prevalence of Haller cell, presence of dehiscence, unilateral, or bilateral frequency were assessed. In addition, the size was categorized in three groups of small, medium, and large. Chi-square and Cochran–Mantel–Haenszel tests were used for statistical analysis of the data, and p < 0.05 was considered to be significant. Results A total of 51 male and 69 female with mean ± SD age of 38.84 ± 68.14 were assessed. The overall prevalence of Haller cells was 56.7%, of which 44 (64.7%) were unilateral and 24 were bilateral (35.3%). The majority of the cells (70.7%) were seen in medium (2–4 mm) sized. There was a significant association between Haller cells and orbital floor dehiscence (p ≤ 0.001). Conclusion The prevalence of Haller cells was remarkably high and the presence of Haller cells was strongly associated with ipsilateral orbital floor dehiscence. Based on the findings of this study, CBCT can be useful in delineation of the bony anatomy of sinonasal complex.


Introduction
One of the problems in oral and maxillofacial procedures is having many diferent anatomical features in diferent patients. Haller cells are one of these normal variations of paranasal and nasal areas, which are related to some symptoms and diseases [1,2]. Tese cells arise from anterior ethmoid air cells and are located in the sinus foor, medial orbital foor, lateral to the maxillary infundibulum, and in the most inferior part of lamina papyracea [1,3]. Haller cells were frst introduced by a Swiss anatomist named "Albert Von Haller" in 1756. Other names for Haller cell are orbitoethmoidal cell and maxilloethmoidal cell [1,4]. In addition to orofacial pain and sinusitis, Haller cells can cause other maladies including nasal congestion, incomplete nasal breathing, headache, chronic coughs, and mucocele [1][2][3].
Haller cell position may lead to disruption of the normal pattern of mucocilliary fow that causes recurrent maxillary sinusitis [4,5]. Tese cells are discovered in paranasal CT examinations by accident [5,6]. Te prevalence of Haller cells in CT examinations has a wide range [1,3,7]. Hui et al. indicated that the prevalence of Haller cell is 29.5% and there was no signifcant relation between the presence of Haller cell and maxillary sinus pathologies [8]. Some studies showed a signifcant relationship between Haller cell size (>3 mm) and orbital foor dehiscence; nevertheless, there is no defnite information on this matter [2,6]. In 2013, Mathew et al. studied CBCTs of 50 patients, which showed a 60% prevalence for Haller cells. Tere was no signifcant relation between the existence and size of Haller cell, size of maxillary ostium and maxillary sinusitis; however, there was a signifcant relation between Haller cell and dehiscence of orbital foor [6]. In 2013, Khayam et al. studied panoramic radiographic images of 200 patients to determine the existence of Haller cells. Te prevalence of Haller cells was 32.5% [9]. In 2012, Raina et al. surveyed panoramic radiography of 600 patients, and the prevalence of Haller cell was 16% [1]. In 2010, Valizadeh also studied 310 panoramic radiographic images, and the prevalence of Haller cells was 37%, which showed that Haller cells can be a common normal landmark in panoramic radiography [10]. In 2005, Lerdlum and Vachiranubhap studied CT images of 133 patients to determine the prevalence of sinus anatomical variations. Haller cell was the second prevalent anatomical variation (9.4%); yet in this study, only agger nasi cells (anterior ethmoidal cells) had a signifcant relation with sinusitis [4]. Haller cells can limit the accessibility to the maxillary sinus and anterior ethmoidal cells in endonasal surgeries; thus, surgeons must be informed about these anatomical variations, which increase the risk of complications after surgeries [1]. Due to the limitations of previous studies and lack of evidence in Haller cell topic, the aim of this study was to determine the prevalence of Haller cells and its relationship with orbital foor dehiscence in CBCT images.

Materials and Methods
Tis cross-sectional study was approved by the Research Committee of Shahid Beheshti University of Medical Sciences and it was conducted in accordance with the Declaration of Helsinki and its subsequent revisions. Te study was conducted in accordance with the STROBE statement. 120 samples were selected by the simple sampling method from referred patients to radiology department of dental school. Te CBCTscans had been requested for purposes not related to this study from 2018 to 2019. Te CBCT scans had been obtained by the New Tom VGI CBCT scanner (Quantitative radiology, Verona, Italy) in two centers with the exposure settings of 110 kVp and 3.3-20 mA. Patients were included if they needed CBCT for diferent purposes (e.g., dental implants, jaw lesions, TMJ, and orthodontic evaluations), and they were excluded if they had a history of tumor, surgery, sinus problems, sinonasal polyposis, trauma of the midface, and if younger than 16. Te images were further analyzed by NNT software in coronal plane by an experienced oral and maxillofacial radiologist.
A meticulous criteria for defning Haller cells as air cells was used, for any size located along medial portion of the orbital foor and/or lamina papyracea inferior to the bulla ethmoidalis and continuous with ethmoid capsula. Te continuity with ethmoid capsula distinguishes Haller cells from infraorbital recess of maxillary sinus. Haller cell size was measured by its maximum mediolateral dimension. Maxillary ostium size is the distance between the most medial part of Haller cell and the uncinate process. Ostium and Haller cells based on size, are divided into 3 groups: small (less than 2 mm), medium (2-4 mm), and large (more than 4 mm). Infraorbital dehiscence is defned as the loss of bone density, and when the diference between dehiscence and thin bone wall is not recognizable, dehiscence diagnosis is acceptable (Figure 1).
All collected data were executed by using the SPSS software (version 19), and the association between them was tested by the chi-squared test and Cochran-Mantel-Haenszel test.

Results
In this study, 120 patients (42.5% male and 57.5% female) were included from ages 18 to 79 with the average of 38.84. Tere was no signifcant relation between Haller cells and gender (p � 0.682). Sixty-eight patients had Haller cells in their CBCTs, so the prevalence of Haller cells in this population was 56.7%, which included 64.7% unilateral and 35.3% bilateral. Both men and women were the same in the number of unilateral Haller cells, but in men, unilateral Haller cells were three times more than bilateral ones. Also, bilateral cases were twice in females. Tere was no signifcant relation between unilateral/bilateral and gender/age (p � 0.186/p � 0.419). Haller cells were categorized into three groups according to mediolateral dimensions: (a) small: less than 2 mm, (b) medium: 2-4 mm, and (c) large: more than 4 mm. Twenty percent of Haller cells were small, 70.67% were medium, and 9.33% were large, so the most prevalent size of Haller cells was medium (p ≤ 0.001). Even though the prevalence of Haller cells is various in diferent ages, we can assume all ages the same (p � 0.282). Furthermore, there is no relation between size of Haller cells and gender/age (p � 0.414/p � 0.668).
From 68 patients having Haller cells, 11 cases had orbital foor dehiscence. Fifty-two participants did not have Haller cells, also did not have orbital foor dehiscence. So, in this study, coexistence of Haller cells and orbital foor dehiscence is confrmed.

Discussion
Tis study estimated the prevalence of Haller cells in CBCT images almost high and about 56.7%. In some studies, an extremely variable range (2%-70.3%) has been reported for the prevalence of Haller cells [8,[11][12][13][14][15][16][17][18][19][20][21][22][23][24][25]. Mathew et al. [6] reported a 60% prevalence for Haller cells and Khojastepour et al. [26] reported 68%, which are close to our study. Alkire and Bhattacharyya [13] also reported a 70.3% prevalence for Haller cells. Tis variability can be due to variation in subjects' race, age, sample size, observer's judgement regarding the presence of Haller cells in images, and diferent defnitions for Haller cells in diferent studies. On the other hand, imaging technique also changes the results. Due to CBCT being a volumetric imaging technique, all the Haller cells in any size get captured; on the contrary, in multislice CT scans, small Haller cells could easily be missed in the interslice intervals [6]. Te high percentage of Haller cells in this study can represent the high sensitivity of CBCT scan in the detection of small delicate bony structures.
In Mathew's study, Haller cells were mostly present bilaterally, which was not statistically signifcant. Te diference in the results can be due to the diference in the population and smaller sample size in Mathew's study (n � 50) than the present study (n � 120). Our study showed that the prevalence of Haller cell in women is slightly higher than in men. However, this diference is not statistically signifcant. Ozcan et al. also indicated that the prevalence of Haller cell is three times higher in female than in male [25]. Tese fndings are consistent with the results of Khojastepour et al. [26], Raina et al. [1], and Basic et al.'s [40] studies.
Te most prevalent Haller cells observed in our study were medium sized (2-4 mm) with a signifcant diference. In the study of Dhillon and Kalra [32], about 51% of cases were large (>4 mm). Te diference between these two studies can be justifed by the variation in the ethnic characteristics of the populations studied and the sample size.
In this study, more than 55% of positive cases were under 40 years old. However, the diference in the distribution of Haller cell prevalence by age was not statistically signifcant. In Raina et al.'s study, 64.6% of Haller cells were observed in subjects from ages 18 to 30, which is consistent with our study results [1].
Various studies have assessed the relationship between presence of Haller cell and maxillary sinus drainage and pathologies such as sinusitis [41][42][43][44]. Although presence of Haller cells may interfere with normal sinus drainage, Suzuki-Yamazaki performed a successful sinus lift procedure in a patient with large Haller cell [41]. To the best of   [45]. Tey declared that any pathologic lesion related to Haller cells should be considered as a potential for unilateral eye cellulitis. Given that there is no lymphatic drainage system in the eye, they propounded a hypothesis that infection spreads through orbital foor dehiscence, lamina papyracea, or sutures of the medial part of orbital foor. Mathew et al.'s study also showed a significant relationship between Haller cells and orbital foor dehiscence; both of these results support our study (3).
So according to this study, we suggest that anytime an infamed Haller cell is observed in CBCT. Simultaneous presence of orbital foor dehiscence should be expected.

Conclusion
Tis study estimated the prevalence of Haller cells in CBCT high (56.7%) and noticeable and also showed that statistically, there is a signifcant relationship between Haller cells and orbital foor dehiscence. It can be concluded that CBCT can be a useful imaging modality for evaluating the anatomical aspect of sinonasal bone complex due to its high accuracy and lower radiation dose.

Limitations and Suggestions
6.1. Limitation. Due to the nature of this study (in vitro) and method of collecting the images (CBCTs stored in university's archive), there was no ethical limitation.

Recommendations.
For increasing the accuracy and efciency of the study, we suggest a larger population with more samples available.
Diferent resolutions should be used in CBCT imaging for better diagnosis of Haller cells and orbital foor dehiscence.

Data Availability
Te datasets used and/or analyzed during the current study are available from the corresponding author upon request.

Ethical Approval
Te study protocol was approved by the Institutional Review Board of Shahid Beheshti University of Medical Sciences.

Consent
All authors gave their consent for the publication of this manuscript.

Disclosure
A preprint has previously been published [46].

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
Mahkameh Moshfeghi designed and supervised the study process. Hamidreza Dehini performed data acquisition and investigation. Ghazizaeh Ahsaie M participated in the data analysis and interpretation. Ghazizadeh Ahsaie drafted the manuscript. All the authors have reviewed and approved the fnal manuscript.