Lateral Cephalograms: Why Do We Take Them?

Common cephalometric analyses

One of the first clinically oriented cephalometric analyses was described by Downs in 1948.⁷ His analysis was based on a group of 20 people with ideal occlusions. Ten cephalometric measurements were described, with the Frankfort plane being the key reference plane. Downs reported notable variation around the mean values for each measure and stated how it was important to use all the measurements together, rather than focus on any single measurement in isolation.

It was not until Steiner in 1960 that the line SN was adopted as the horizontal reference plane.⁸ The Steiner analysis however, is reportedly based on his cephalometric measurement of just two Hollywood stars.

Probably the most popular analysis used in the UK is the Eastman analysis, based on the work of Ballard who analysed 250 London school children of various ages,⁹ and which was subsequently refined by Mills.¹⁰ Commonly used cephalometric points, planes and angles are depicted in Figure 1, with Eastman standard norms and abbreviations listed in Table 1. Cephalometric assessment of skeletal pattern can be determined by comparing the relationship of the maxilla and mandible with the cranial base using angles SNA and SNB. The ANB angle, the difference between SNA and SNB, classifies the skeletal pattern (ANB <2° =Class 3; ANB >4° = Class 2; ANB between 2° and 4° = Class 1).

Cephalometric norms have often been based on small sample sizes and the scientific basis of analyses based on these norms is therefore sometimes weak. In 1956, Graber described some of the limitations of cephalometrics, in particular criticizing the reliance on arbitrary norms and stating that mathematical descriptions of morphology are likely to be an oversimplification.¹¹ For example, when analysing the anteroposterior position of the jaws, there are problems associated with relying only on the ANB angle.¹² The Eastman correction described by Mills (1970) attempts to correct for an aberrant position of nasion when SNA is too high or too low and the SN/MxP angle is normal.¹³ However, the Eastman correction is itself subject to error because in a study based on a geometrical model, the Eastman correction was shown to overestimate towards the opposite skeletal discrepancy when N was moved posteriorly towards S (ie direct measurements showed a Class 2 tendency, whereas the Eastman correction revealed a Class 3 tendency).¹⁴ An additional analysis such as the Wits appraisal, which relates anteroposterior discrepancy of the jaws to each other and not to the cranial base, is also recommended.¹⁵ This technique however, was only based on 21 male and 25 female subjects, and relies on correctly identifying the functional occlusal plane, the orientation of which changes with growth or treatment.

What are the errors in cephalometry?

Considerable error can be associated with cephalometric measurements, with two general categories of error being described.¹⁶ The first category is errors of projection, relating to the radiograph being a 2D shadow of a 3D object, with non-parallel X-rays producing the shadow. These radiographs are always therefore distorted enlargements. The magnitude of the enlargement is related to distances between the focus of the X-ray beam, the head and the film. Objects close to the film and in the centre of the X-ray beam show the least magnification (Figure 2). As such, any change in orientation of the head away from natural head posture will lead to a corresponding distortion of the image.

The second category is errors of identification. Landmark identification using points is associated with an ‘envelope of error’ (Figure 3). The magnitude of the error varies greatly from landmark to landmark and each landmark point has its own, usually non-circular, envelope of error. Sella has been identified as the most reliable point, because identification of the centre of a spherical shape is straightforward.¹⁷ Identifying the shallowest aspect of a gradual curve (eg A point or B point) is more prone to error compared to identifying an edge where two well-defined planes intersect (eg upper incisor edge)¹⁶ or a defined suture, for example the naso-frontal suture. In addition, the image resolution and contrast also affect point identification, as will X-ray image orientation.

Angular and linear measurements are also necessarily affected by both projection and landmark identification errors. For linear measurements, the shorter the distance between two points, the greater the percentage error introduced by a measurement error of a given size. Different angular and linear measurements have different levels of reliability, and this is of critical importance in cephalometric superimposition. Baumrind and Frantz calculated that an observed ANB change of more than 1.12 degrees is required to be considered clinically significant, and not just the effect of error.¹⁸ As described in Figure 4, use of a horizontal reference plane is important.

The introduction and use of 3D cephalometry has increased in recent years, with techniques including CBCT,¹⁹ low-dose multi-slice CT or MRI. 3D imaging (Figure 5) may be particularly helpful for assessing facial asymmetry, due to better representation of the real morphology of the skull, unlike in traditional lateral cephalometry, where structures on the left and right are superimposed on each other.²⁰ It is thought that 3D landmark identification and measurements are at least as reliable as traditional 2D cephalometric techniques.²⁰

Disadvantages of conventional radiography and CBCT include the radiation doses, which are often applied to young patients. In a child, the diagnostic reference levels (X-ray energy delivered to the patient) of a lateral cephalogram are about 34 times that of a typical peri-apical radiograph, whereas CBCT of an impacted maxillary canine is 230 times that of a typical peri-apical.²¹ MRI techniques, which do not use ionizing radiation, may provide a solution for the future. There is evidence that MRI (Figure 6) may provide good levels of diagnostic and therapeutic efficacy equivalent to traditional cephalometry, despite the fact that MRI is principally used in soft tissue diagnostics.^22,23 Artefacts from metallic objects remain a limitation to this technique, which would be particularly relevant for mid-treatment cephalometric assessments.²² The lack of radiation associated with MRI would, however, facilitate justification of post-treatment imaging.

To date, there is limited research-based evidence on the use of 3D cephalometry²⁰ and, as with 2D techniques, true diagnostic efficacy is difficult to measure because it is impossible to check real landmark positions in patients. Since studies on patients can only describe ‘observer performance’ or ‘reliability of the methods’,²⁰ cadaver studies or other in vitro methods might be used as an alternative means to identifying potential errors.

The Michigan growth study

This study started in 1935 and lateral cephalograms of subjects were taken routinely until about 1970. The study resulted in publication of an atlas of craniofacial growth with cephalometric standards.³⁴

The Burlington growth study

This project started in the Canadian town of Burlington in 1952 and 1258 children were enrolled, which represented nearly 90% of the children within the specified ages required. The population of the town in 1952 was 9000. For the serial experimental sample, six lateral cephalograms and one hand–wrist radiograph were taken between the ages of 3 and 21 years.³⁵

The Bolton–Brush study

The collection contains data from around 6000 research subjects from two large studies (Bolton and Brush). The Bolton study was initiated by B. Holly Broadbent Sr (the father of modern cephalometry) in the 1920s and comprised annual cephalometric radiographs, alongside numerous other records, of subjects throughout babyhood, childhood and adolescence. Data continued to be gathered until 1959.³⁶ The Brush inquiry ceased in 1942 and included 250,000 radiographs of 4000 children. More recent recalls of participants from these studies have confirmed that craniofacial growth continues throughout adulthood.³⁷ More than 100 individuals were recalled in the early 1980s for new radiographs. There were increases in all of the facial dimensions, but vertical changes were the most prominent. Growth rotations of the jaws also continued into adult life.

These growth studies, and others, provide a wealth of data allowing treatment comparisons with untreated controls, matched for age and gender.³⁸ It is highly unlikely that ethical approval would be obtained today for similar large-scale studies with repeated radiographs of healthy subjects.

Bjorks' structural signs

In the 1960s, Björk performed experiments whereby tantalum implants, measuring 1.5 mm in length and 0.5 mm in diameter, were placed subperiosteally to serve as fixed reference points in serial lateral cephalograms.³⁹ Using this technique, Björk reported on seven signs visible on a lateral cephalogram (Figure 13) that can be used to help predict the direction of mandibular growth with particular reference to growth rotations.⁴⁰ Forward rotation was found to be more common than backward rotation. Predicting the growth pattern prior to treatment may be very helpful. Extreme rotation may influence the path of eruption of teeth.⁴⁰ In addition, Class II functional treatment of hyperdivergent patients may be more challenging than hypodivergent, as the backwards growth rotation may displace the chin downwards and backwards, negating the effects of the functional appliance.⁴¹ Unfortunately, evidence suggests that hypo- or hyperdivergent morphology cannot be reliably detected using Björk's structural signs of mandibular growth rotation. Although subjects with the extremes of hypo- or hyperdivergency were all correctly identified using this method in a study in 2005, no statistical association was found between the structural signs and type of growth rotation.⁴²

Cervical vertebral maturation

More recently, lateral cephalograms have been used to assess the timing of growth by measuring maturity of the cervical vertebrae (Figure 14). This method, first described by Lamparski⁴³ and later refined by Baccetti, enables prediction of peak mandibular growth through analysis of the vertebrae C2, C3 and C4.⁴⁴ This cervical vertebral maturation (CVM) technique has found favour over other radiographic methods, such as hand–wrist radiographs, because lateral cephalograms are already commonly prescribed in orthodontics and hence, radiation doses are limited. Hand–wrist radiograph assessment of skeletal maturity is the gold standard, but the CVM method is said to show a high level of correlation with hand–wrist radiographs,⁴⁵ and is both reproducible and reliable.⁴⁶ Some authors, however, have recently challenged the reported reliability of the CVM method,^47,48,49 suggesting reproducibility is poor and that additional biological indicators need to be used. This is likely to be due to the subjectivity associated with assessment of shape changes.