Morphology of the Sella Turcica: A Meta-Analysis Based on the Results of 18,364 Patients

Background: This meta-analysis aimed to present data on the sella turcica (ST) morphology and variations. Furthermore, a detailed morphometric analysis of the ST was conducted. Methods: Major online databases such as PubMed, Scopus, Embase, Web of Science, and the Cochrane Library were searched through. The overall search process was conducted in three stages. Results: This meta-analysis was based on the results of 18,364 patients and demonstrates the most up-to-date and relevant data regarding the morphology of the ST in the available literature. Four classification methods of the ST shape can be distinguished, in which the most commonly occurring variants are the normal ST (55.56%), the U-shaped ST (73.58%), the circular type of ST (42.29%), and non-bridging ST (55.64%). The overall midpoint height of the ST was 6.59 mm (SE = 0.13). The overall length of the ST was 9.06 mm (SE = 0.15). The overall volume of the ST was 845.80 mm3 (SE = 288.92). Four main classification methods of ST morphology can be distinguished in the available literature. Various morphometric characteristics of the ST may be applied in clinical practice to evaluate its shape, dimensions, and normal or pathological variants.


Introduction
The sella turcica (ST) is a saddle-shaped depression on the superior part of the sphenoid bone ( Figure 1) [1]. It is located in the middle cranial fossa, and it is bounded posteriorly by the dorsum sellae (DS) and anteriorly by the tuberculum sellae (TS) [2][3][4][5][6][7][8]. The hypophyseal fossa is the most inferior part of ST (Figure 1), and it contains the hypophysis, also known as the pituitary gland (PG). The PG consists of three lobes: the anterior lobe (adenohypophysis), the intermediate lobe (a thin layer of cells), and the posterior lobe (neurohypophysis) [9][10][11]. It is one of the endocrine system's key regulatory structures since it secretes numerous hormones that influence other endocrine glands [12].
Normality in anatomy can be considered a type of arrangement based on numerous repeated observations [15]. Various studies were conducted to assess ST morphology, yet most included a small number of participants and obtained different results. Thus, the main objective of the present meta-analysis was to provide reliable and objective data on the structure of ST in various patient groups measured using computed tomography (CT) or cephalograms (CGs). This meta-analysis provides clinically useful information on ST morphology.  Due to anatomical relationships, the alteration in ST shape and dimensions, e.g., ST bridging, can be clinically significant. For example, deviations in the ST morphology are mostly seen in patients with various congenital disorders such as Down syndrome, Seckel syndrome, Williams syndrome, and Axenfeld-Rieger syndrome. Furthermore, this causation can be reversed, and PG ailments can cause ST malformations [12][13][14]. It can be a significant factor to consider, as studies have shown that calcification of the ST ceases after full PG development. Therefore, pathologies causing PG enlargements, such as adenomas or cysts, or shrinkage, e.g., primary hypopituitarism or growth hormone secretion deficiency, may potentially impact the ST's definitive shape [12,14].
ST and PG have two embryological origins, with the posterior part of each structure derived from the paraxial mesoderm, which is highly dependent on notochordal induction, and the anterior part originating from the neural crest cells. This factor, along with the different embryogenic processes, explains the observation that anomalies of the anterior wall of the ST appear to be associated with changes in the frontonasal area and body axis defects. In contrast, anomalies of the posterior wall of the ST are linked to changes in the cerebrum [12]. Hence, the ST size and shape are closely related to PG morphology.
Normality in anatomy can be considered a type of arrangement based on numerous repeated observations [15]. Various studies were conducted to assess ST morphology, yet most included a small number of participants and obtained different results. Thus, the main objective of the present meta-analysis was to provide reliable and objective data on the structure of ST in various patient groups measured using computed tomography (CT) or cephalograms (CGs). This meta-analysis provides clinically useful information on ST morphology.

Search Strategy
For the sake of this meta-analysis, a systematic search involved all articles in which the morphology of the ST was evaluated. Major online databases such as PubMed, Scopus, Embase, Web of Science, and the Cochrane Library were searched through. The overall search process was conducted in three stages. (1) In the first step, all mentioned medical databases were searched using the following search terms: [(sella turcica) OR (turkish saddle) OR (hypophyseal fossa) OR (tuberculum sellae) OR (dorsum sellae)] AND [(anatomy) OR (topography) OR (variation) OR (structure) OR (size) OR (morphology) OR (width) OR (length) OR (type)]. No date, language, article type, or text availability conditions were applied. (2) Furthermore, the databases were searched through once again using another set of search phrases: (a) ( . Additionally, each phrase was checked for differences in the structure of words (e.g., variation vs variations) and adapted if needed. (3) A manual search was also performed throughout all references from the initial submitted studies. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed. Additionally, The Critical Appraisal Tool for Anatomical Meta-analysis (CATAM) and Anatomical Quality Assessment Tool (AQUA) were used to provide the highest-quality findings [16,17].

Eligibility Assessment and Data Extraction
The inclusion criteria were set as follows: original articles with extractable data on the anatomy of the ST. The exclusion criteria involved conference reports, case reports, case series, reviews, letters to the editor, and studies with no relevant or incompatible data. Two independent researchers performed a systematic search. A total of 15,752 articles were initially found. After the removal of duplicates and irrelevant reports, a total of 91 articles matched the required criteria and were taken into consideration in this metaanalysis [1,2,9,[12][13][14]. The overall process of article collection is shown in Figure 2. Moreover, the characteristics of the submitted studies are presented in Table 1. Table 1. Characteristics of the studies included in this meta-analysis.

First Author
Year Continent Country Method n  Two independent researchers extracted data from qualified studies. Qualitative data were collected, such as the year of publication, country, and continent. Furthermore, quantitative data were gathered in several categories:

Statistical Analysis
To perform this meta-analysis, STATISTICA version 13.1 software (StatSoft Inc., Tulsa, OK, USA), MetaXL version 5.3 software (EpiGear International Pty Ltd., Wilston, Queensland, Australia), and Comprehensive Meta-analysis version 4.0 software (Biostat Inc., Englewood, NJ, USA) were applied. A random effects model was used. The Chi-square test and the I-squared statistic were chosen to assess the heterogeneity among the studies [103,104]. The p-values and confidence intervals were used to determine the statistical significance between the studies. A p-value lower than 0.05 was considered statistically significant. The differences were considered statistically insignificant in the event of overlapping confidence intervals. I-squared statistics were interpreted as follows: values of 0-40% were considered as "might not be important", values of 30-60% were considered as "might indicate moderate heterogeneity", values of 50-90% were considered as "may indicate substantial heterogeneity", and values of 75-100% were considered as "may indicate substantial heterogeneity".

Results
This meta-analysis was finally based on the results of 18,364 patients. The prevalence of different types of ST was estimated using four classification methods.
The first classification method was general classification ( Table 2). In this categorizing mode, the most prevalent ST type was the normal ST, with a pooled prevalence of 55.56% (95% CI: 49.32-61.71%). "Normal ST" was classified as ST without sella turcica bridges, irregularities of the posterior part of the dorsum sella, oblique anterior wall, the double contour of the floor, or a pyramidal shape of dorsum sellae. The sella turcica bridging is an anatomical variant resulting from the partial or complete fusion of the anterior and posterior clinoid processes. Among reports using this classification, the sellar bridges were observed in 11.34% (95% CI: 8.21-14.91%). The results mentioned above were based on a total of 5406 patients.
In the second classification method (distinguishing U-shaped sella turcica, J-shaped sella turcica, and flat-shaped sella turcica on the saggital sellae section; Table 2), the most prevalent ST type was the U-shaped ST, with a pooled prevalence established at 73.58% (95% CI: 54.24-89.36%). Furthermore, J-shaped ST was found in 16.91% (95% CI: 10.29-24.72%). Those results were based on a total of 1511 patients.
In the third classification method (recognizing circular type sella turcica, oval type sella turcica, and flat type sella turcica; Table 2 and Figure 4), the most prevalent ST type was the circular type of ST, with a pooled prevalence established at 42.29% (95% CI: 24.84-60.73%). Furthermore, the oval type of ST was present in 33.90% (95% CI: 21.10-47.97%). Those results were based on a total of 1191 patients.
In the fourth classification method, classifying types of sellar bridges (Table 2), the most common ST type was the type I, i.e., an ST without bridging, with a pooled prevalence of 55.64% (95% CI: 44.33-66.66%). Furthermore, type II, an ST with partial bridging, was found in 32.99% (95% CI: 24.22-42.39%). Those results were based on a total of 1791 patients. All of the results above and more detailed data regarding the types of ST are demonstrated in Table 2.
The main important ST morphometric characteristics are as follows. The overall anterior height of the ST was found to be 6.71 mm (SE = 0.52), whereas the midpoint and posterior heights of the ST were found to be 6.59 mm (SE = 0.13) and 6.93 mm (SE = 0.22), respectively. The detailed results regarding the height of the ST are presented in Table 3. The overall length of the ST was found to be 9.06 mm (SE = 0.15). In females, the ST length was 8.

Discussion
Three main classification systems that describe the morphology of the ST can be distinguished. The first one is derived from the study of Axelsson et al. [102], where seven ST types were distinguished. Cited authors applied the term "normal ST morphology" described by Björk and Skieller in 1983 [105]. That way, the classification was created, which divides the ST into normal ST, oblique anterior wall, sella turcica bridge (STB), double contour of the sella turcica floor, irregularities of the posterior part of DS, pyramidal shape of DS, and variants showing characteristics of more than one type. The prevalence of each type mentioned above varies wildly in the literature, with the normally shaped sella turcica present in most patients. For instance, Islam et al. [2] assessed the morphologic properties of the ST using CT images gathered from 166 patients. Those authors observed normal sella in 69.2% of subjects, while Isman et al. [1], using cone beam computed tomography (CBCT), described it in a remarkably lower percentage of patients, with only 49.8% having an ST of this kind. However, the current meta-analysis's findings indicate that 55.56% of individuals have this form of ST. The prevalence and morphology of the second most common type varied between the studies. Islam et al. described irregularity (notching) in the posterior part of the dorsum sella in 16.2% of patients [2]. In comparison, Isman et al. reported a double contour of the sella turcica floor with a prevalence of 22.8% [1]. In contrast to those results, our study concluded that the sella turcica bridge (STB) was the second most common type, with a prevalence of 11.34%. Interestingly, in both studies mentioned above, the frequency of this morphologic type was on the lower end of the spectrum, with 0% and 3% of subjects presenting it, respectively [1,2].
It is crucial to note that some authors further divided ST using the classification by Leonardi et al. [106]. Those authors categorized ST into three groups based on the formation of the STB: no bridging (type I), partial bridging (type II), and complete calcification of the ST (type III) [106]. The assessment was based on comparing the length of ST and the anteroposterior greatest diameter. The ST length was measured as the distance between the tuberculum sellae and the tip of the dorsum sellae, and the anteroposterior greatest diameter was measured between the tuberculum sellae and the furthest point on the interior wall of the pituitary fossa. When the length was greater or equal to three-quarters of the diameter, ST was classified as type I; if this length was lesser, type II was assigned; and if there was a visible diaphragm sella (complete calcification of the interclinoid ligament), it was defined as type III. Notably, this method was only used in studies assessing ST with CGs. Interestingly, most studies reported type I with the highest prevalence, although the exact percentage varied across analyzed papers [45,55,87]. The results of the present metaanalysis align with the literature data, with 55.64% frequency for type I. It is also worth mentioning that the forming sellar bridges in the ossification stage were observed even during the fetal period (one observation of the fetal skull), suggesting that this type of ST formation may also have a developmental background, contrary to age-related calcifications occurring in some cases [107].
Another classification system that presents the morphological pattern of the ST is based on the resemblance of the letters U and J, as proposed by Ruiz et al. (2008) [74]. The U-shaped ST is distinguished when the sellar tubercle (tuberculum sellae) and the dorsum sellae (DS) are maintained at the same height. The J-shaped ST is described when the sellar tubercle is positioned inferior to the DS. The third type appearing in this classification, the flat-shaped (shallow) ST, is described when the depth of ST is minimal [66]. Hasan et al. [66] analyzed CT scans of 71 individuals and found that 50.7% had a U-shaped ST, 32.4% had a J-shaped ST, and 16.9% presented with a flat-shaped ST. However, a study by Muhammed et al. [54] concerning differences in ST morphology between Bosnian and Iraqi populations concluded that in both groups, the prevalence of U-shaped was much higher, with a prevalence of 86.7%. Accordingly, the frequency of other types was curtailed, with 12.2% and 9.4% possessing the J-shaped ST and 1.1% and 3.9% possessing the flat-shaped ST in both populations, respectively [54]. The results of our meta-analysis indicate that the distribution of each type of ST differs from the above papers. Nevertheless, the U-shaped ST is still the most prevalent, with 73.58% of individuals possessing it.
Another less frequently used classification system dating back to 1923 was provided by Gordon et al. [2,31,95,108]. This classification categorizes the ST as circular, oval, or flat-shaped. Most past research demonstrated the oval type as the most prevalent. However, in two separate articles conducted by Yasa Y. et al. [31,32], the most frequent types of ST were stated to be circular, seen in 65.3% and 69.5%. Our statistical analysis revealed a significantly lower percentage of individuals classified as having the circular type of ST, specifically 42.29%. Conversely, the oval type was observed in 33.9% of the subjects.
The morphometric properties of the ST have been widely discussed in the literature. The heights of ST were measured as a distance between the sella floor and certain structures perpendicular to the Frankfort horizontal plane [109]. This plane is described as a line connecting the left orbital and both Porion points, the most superiorly positioned points of each external auditory meatus [109,110]. To estimate the anterior height, ST was used as a landmark; to estimate the posterior height, the posterior clinoid process (PClin) was chosen; and to estimate the median height, a point midway between TS and PClin was used [22,28,83,88]. When analyzing the results obtained in our meta-analysis, both anterior and posterior ST heights varied considerably between the two radiological modalities, i.e., CT and CG. Both measurements were significantly lower in patients evaluated by CT; however, the standard error was much higher, suggesting that the sample of cases examined by this method was too small. Interestingly, the dimensions established in patients with a cleft lip and palate were decreased compared to the overall results.
Moreover, the results of the present meta-analysis show the ST length, measured as a distance from TS to PClin [22], to be the longest in the North American population and the shortest among Europeans. It is worth noting that the standard error was significantly higher in the former group, most likely because of the scarcity of studies. Hence, the results may not be fully representative of that population. The diameter of ST varied considerably in the literature, from 9.4 mm [69] to 14.24 mm [46]. Our meta-analysis concluded it to be 11.15 cm, without special deviations between Europeans and Asians or differences resulting from using various radiological methods. However, we noticed a slightly smaller diameter of ST in subjects with skeletal Class II compared to other skeletal classes. This observation was, to some extent, expected, looking at previous studies [26,39,78]. Nevertheless, some previous papers comparing measurements between different skeletal classes described the diameter of the ST as being higher in skeletal Class II compared to skeletal Class I or even found it to be the biggest among all skeletal classes [21]. On top of that, the mean ST diameter in patients with Down Syndrome was close to that of healthy individuals; however, further studies are needed to evaluate those calculations. Interestingly, the ST area, i.e., the area outlined by the sella contour and line joining TS and PClin [22,83], was calculated to be much smaller in patients with any type of cleft lip and palate compared to healthy individuals (43.12 mm 2 vs. 56.64 mm 2 ). In addition, the ST area was higher in Asians compared to the European population (62.32 mm 2 vs. 56.29 mm 2 ).
The mean ST depth varied significantly in the literature from 6.4 mm [39] to 10.87 mm [23]. The results of the present meta-analysis show that the overall depth of the ST is 8.00 mm, with minor differences between the analyzed groups. In addition, the ST depth was smallest in patients with skeletal Class II, which aligns with most studies [26,39,78]. The literature provides differences in ST volume ranging from 340.5 mm 3 , described by Halit et al. (2014), to 1428 mm 3 , described by Singhellakis et al. (1983) [37,75]. In our meta-analysis, the mean ST volume was 980.75 mm 3 . Interestingly, the overall calculated ST volume was smaller (845.8 mm 3 ), but the standard deviation was high; hence the values varied greatly between individuals.
As the pituitary gland is located within the pituitary fossa and markedly influences the size of ST [12], any deviation in its measurements may reflect possible pathologies, even before they become clinically evident [83]. While CGs are not considered the method of choice for diagnosing pituitary tumors, the presented data can be used to avoid overlooking any incidental findings [83]. Therefore, up-to-date data concerning the ST's morphology can help clinicians more accurately spot abnormalities in the PG size assessed using CGs and more sensitive methods, such as CT. Moreover, this knowledge is crucial for neurosurgeons performing surgery on the cranial base to choose the most suitable technique and spare damage to the pituitary gland and adjacent structures [49]. On top of that, more accurate calculations are also crucial in orthodontics, as ST is an essential anatomical determinant utilized to diagnose maxillofacial disharmonies such as palatally displaced canines or the congenital absence of the second mandibular premolar [12,55,106]. It can also help assess the results of orthodontic treatment [12]. The ST is also used as a landmark in lateral cephalometric analysis to assess the jaw relationship [49].

Study Limitations
The present meta-analysis has several limitations. Due to the nature of the systematic review, the accuracy of the final results depends on the quality of the primary studies. A few limitations must be taken into account regarding this issue. First, the term population is often overused in research and refers to a sample that may only partially reflect the characteristics of the general population. It should also be considered that the samples used in the classification of U, J, and flat-shaped ST mainly consist of Bosnian, Chinese, Nepalese, and Iraqi individuals; therefore, the data should be related to other ethnicities with great caution. Most STs were also studied in Asia (n = 8791). Therefore, the overall results of this meta-analysis may be burdened, as they may reflect the anatomical features of Asian individuals rather than the global population.
Additionally, some analyses concerning ST sexual dimorphism or the discrepancies between the results obtained with different measurement methods (or various diagnostic modalities) were not performed due to insufficient literature data or the significant risk of possible results bias. Another limitation is that the article is based on a mix of radiological imaging rather than anatomical studies involving dry skulls or wet specimens. In addition, data on internal carotid artery anatomical variations or anomalies in the vicinity of sella turcica were not included in the study. As described by Björk and Skieller [105], "During growth, sella turcica increases in size by apposition on tuberculum sellae and by resorption at the posterior wall and at the floor". Thus, the sella turcica's age-related changes should also be considered when assessing the sella turcica contour morphology and comparing various examined samples. Regardless of limitations, our meta-analysis attempts to establish the detailed morphology of the ST based on the data from the currently available literature, which meets the requirements of evidence-based anatomy.

Conclusions
Four main classification methods of ST morphology can be distinguished in the available literature. Various morphometric characteristics of the ST may be applied in clinical practice to evaluate its shape, dimensions, and normal or pathological variants.

Conflicts of Interest:
The authors declare no conflict of interest.