A indicates glabellar; B, piriform; and C, maxillary angle.
A indicates anterior lacrimal crest; B, orbital; C, zygoma; D, piriform. All measurements are 2-dimensional.
eFigure 1. Example Measurements of Glabellar Angle
eFigure 2. Measurement of Piriform Width at Its Widest Point
Customize your JAMA Network experience by selecting one or more topics from the list below.
Paskhover B, Durand D, Kamen E, Gordon NA. Patterns of Change in Facial Skeletal Aging. JAMA Facial Plast Surg. 2017;19(5):413–417. doi:10.1001/jamafacial.2017.0743
What are the patterns of bony changes in the aging face when patients are observed longitudinally?
In this case series study of 14 adults, a significant decrease in both maxillary and piriform angles was found on 3-dimensional analysis of computed tomographic scans repeated at least 8 years apart. Two-dimensional analysis revealed significant decreases in glabellar and maxillary angles and significant increase in piriform width, as well as differences between sexes.
A longitudinal design to study facial aging as well as a 2-dimensional method of analysis can improve on previous methods of characterizing the bony changes that occur in the aging face.
Research in facial aging has focused on soft-tissue changes rather than bony changes despite evidence of the importance of underlying bony structural changes. Research has also been limited by comparing different patients in separate age groups rather than the same patients over time.
To longitudinally document patterns of change in the facial skeleton and determine a consistent methodology for measuring these changes.
Design, Setting, and Participants
Case series study of university hospital system records using facial computed tomographic (CT) images timed at least 8 years apart in adults initially aged 40 to 55 years with no history of facial surgery who required repeated facial imaging that included the entire midface and cranium.
Main Outcomes and Measures
Face CTs were analyzed for 3-dimensional constructions and 2-dimensional measurements to document changes in glabellar, piriform, and maxillary angles and piriform height and width.
Fourteen patients (5 men, 9 women; mean [SD] age, 51.1 [5.8] years) with mean (SD) follow-up of 9.7 (1.4) years were eligible for 2-dimensional analysis, which revealed statistically significant decreases in mean (SD) glabellar angles (from 68.8° [7.6°] to 66.5° [8.6°]) and maxillary angles on both the right (from 82.5° [6.3°] to 81.0° [7.1°]) and left (from 83.0° [5.8°] to 81.0° [7.0°]), as well as increases in mean (SD) piriform width (from 24.5 [1.6] mm to 25.5 [1.3] mm). Nine patients (3 men, 6 women; mean [SD] age, 51.4 [6.3] years) with mean (SD) follow-up imaging at 9.6 (1.5) years were eligible for 3-dimensional analysis, which revealed statistically significant decreases in mean (SD) maxillary angles (from 56.5° [6.6°] to 51.6° [7.6°]) and piriform angles (from 50.8° [3.4°] to 49.1° [3.4°]). Statistically significant differences between the sexes were also noted: Initial mean (SD) glabellar angle for men was 61.7° (5.7°) vs 72.7° (5.4°) for women, with final values of 57.9° (4.9°) vs 71.2° (6.0°). Mean (SD) maxillary angle initial values were 87.8° (6.1°) (right) and 87.1° (4.9°) (left) for men, with 79.6° (4.3°) and 80.6° (5.0°) for women, respectively. Final values were 87.0° (4.4°) and 86.9° (4.1°) for men and 77.7° (6.1°) and 77.7° (6.2°) for women, respectively. Mean (SD) piriform height for men was 35.0 (2.0) mm initially and 35.5 (2.1) mm finally, vs 31.3 (2.8) and 31.6 (3.0) mm for women, respectively.
Conclusions and Relevance
Our pilot study of repeated CT images of patients over several years supports previous studies of bony facial aging and further characterizes these changes. This study is the first, to our knowledge, to document bony changes of the face in the same group of patients at different time points to better characterize facial aging. We also detail an improved methodology to study bony aging to contribute to additional research in the field.
Level of Evidence
For more than half a century, the medical literature has shown that all 4 structural tissue types—skin, fat, muscle, and bone—contribute to facial aging.1 However, the extent to which each of these tissues contributes to the facial aging process continues to be debated.
During this time, it has been clear that the majority of facial aging is due to gravitational effects on the facial soft tissues, validating the “surgical model” in which lifting, repositioning, and excising excess soft tissue is the optimal solution to restore the face to its prior form. There are, however, inadequacies in the aesthetic outcomes, especially when techniques that require skin tension are applied. Limitations are also apparent in the extremes of age and in specific anatomic locations such as the periorbital and nasolabial regions.
Volume restoration has also been advocated as a technique for facial rejuvenation during this period,2,3 but it was the apparent inability of the “surgical model” to consistently address certain facial aging patterns that led to a new emphasis on understanding the contribution of volume loss and the recent popularity of the “volume model.” This model emphasizes deflation and touts the resultant folds and hollows as evidence of facial fat loss causing the majority of aging changes. Advances in soft-tissue fillers and fat transfer techniques, as well as the ease, cost, and consumer demand for the “volume model,” have fueled increased use of this approach despite similar evidence of limitations and inadequacies in aesthetic outcomes. In addition, to our knowledge there has been no study documenting the loss of facial fat during the aging process.
Starting with studies from Pessa and Lambros, research in the early 2000s documented the neglected contribution of the facial skeleton in elucidating the aging process.1 Of interest, skeletal changes occur in the exact regions where volume enhancements have their greatest success: the medial cheek and periorbital region. In addition, bone loss also explains certain limitations to the surgical model approach to the aging face, especially in the extremes of age.
In his work, Pessa1 discussed Lambros’s theory of bony facial aging and verified it with the use of 3-dimensional (3-D) stereolithography in 12 male participants separated into 2 groups by age. The study supported the theory that midface changes can be thought of as a clockwise rotation in the sagittal plane, relative to the skull base, with maxillary retraction and increasing prominence of the glabellar angle. Measurements were performed on 3-D stereolithographs with a ±1-mm variability during the printing process, and the axial computed topography (CT) data that were used for the rapid prototyping were noted to be up to 3-mm thickness. The combination of the variability during the printing process and the rather large gaps between axial data brings to light the difficulty in accurately measuring the changes involved.
Nearly a decade later, Shaw and Kahn2 published work on the aging face and its aesthetic implications with the use of CT and 3-D volume rendering. Using 1-mm slice thickness, digital measurements were performed on the Volume Viewer (Voxtool 3.0.64q; GE Healthcare) platform. Using 3 separate patient groups of young, middle-aged, and old, they were able to once again denote the various bony morphologic changes including increases in orbital aperture width, orbital area, and piriform aperture area. They also noted a significant reduction in glabellar and maxillary angles. Although the study clarified the changes that occur to the bony skeleton, it was inherently limited by the use of separate patient groups.
Our goal in this pilot study was to identify methodology to accomplish a longitudinal study to characterize the timing and pattern of these bony changes to define the contribution of facial skeletal bone loss and remodeling to benefit our general understanding of the aging face. Methodology to study patients longitudinally is necessary to investigate any interventions aimed at treating, slowing, or preventing these bony changes. Our present study is the first to identify these morphological changes using improved methodologies while following individual patients longitudinally.
A retrospective query of our institution’s radiology report database (Montage Healthcare Solutions) was performed to identify adult patients initially aged 40 to 55 years with repeated CT studies at least 8 years apart that included the entire midface and cranium. This study was approved by the Yale School of Medicine institutional review board, and a waiver for informed consent was granted due to its retrospective nature and use of deidentified patient information. A total of 14 patients meeting criteria were selected for further review, with the majority having undergone CT angiography surveillance for intracranial aneurysm growth. Patients were excluded from study participation for evidence of prior significant calvarial or facial surgery, noted current or previous facial trauma, abnormal-appearing soft-tissue or bony structures consistent with pathologic disease processes, or complete lack of maxillary dentition. Slice thickness ranged from 0.63 to 1.25 mm.
Three-dimensional reconstructions were performed on a dedicated GE Workstation. The skull was positioned so that the sella-nasion line was oriented parallel to the axial plane. Anthropometric measurements obtained were glabellar angle, piriform angle, and maxillary angle according to previous authors (Figure 1).1,2 Piriform distance, anterior lacrimal crest distance, and orbital and zygoma distance were measured from a line drawn inferiorly from the nasofrontal suture, with measurements parallel to the nasion-sella line similar to previous authors (Figure 2).3 Using the GE 3-D volume-rendering software, we were able to 3-D reconstruct 9 of the 14 patients’ initial images. Due to limitations with software compatibility and the images’ raw data, we were not able to reconstruct 5 of our patients’ images using the 3-D software. This led us to establishing a methodology for 2-dimensional (2-D) analysis of the aging face.
Two-dimensional multiplanar reformatted images were reconstructed using Vitrea software (Vital Images) on all 14 patients. The following measurements were obtained on the midsagittal slice: glabellar angle (eFigure 1 in the Supplement) (apex anterior at the glabella with one side through the nasofrontal suture and the other side parallel to the sella-nasion line) and piriform height (measured from rhinion to nasal crest). The piriform width was obtained on coronal reformatted images and defined as the maximal transverse width of the piriform aperture (eFigure 2 in the Supplement). The left and right maxillary angles were obtained on parasagittal slices at the level of the infraorbital foramen and mid-orbits (apex anterior at the superior maxilla with one side through the mid-sella and the other side to the inferior maxilla–alveolar crest junction) (Figure 3). Of note, the superior line universally was parallel and approximated the orbital floor, similar to previous authors’ methods.4
Statistical analysis was performed using Microsoft Excel, version 14.4.7. Absolute change in anthropometric parameters and mean values between time points were calculated for each patient. Paired and unpaired t tests were performed when appropriate. The correlation coefficient between 3-D and 2-D measurements was obtained as well when appropriate. P ≤ .05 was considered statistically significant.
Our 3-D analysis of facial aging included 9 patients (3 men and 6 women), with a mean (SD) starting age of 51.4 (6.3) years and follow-up of 9.6 (1.5) years. We noted a statistically significant decrease in the mean (SD) maxillary angles from 56.5° (6.6°) to 51.6° (7.6°) (P = .001) and a decrease in piriform angle from 50.8° (3.7°) to 49.1° (3.4°) (P = .02). Piriform distance, anterior lacrimal crest distance, and orbital and zygoma distance results noted no statistically significant difference between the initial and final images (Table 1).
Fourteen patients (5 men and 9 women) were included in our 2-D analysis of facial aging. The mean (SD) age was 51.1 (5.8) years at initial imaging, with changes of the initial and final images documented over a mean (SD) 9.7 (1.4)-year time frame (Table 1). We noted statistically significant differences in the mean glabellar angle (from 68.8° [7.6°] to 66.5° [8.6°]), both right and left maxillary angles (from 82.5° [6.3°] to 81.0° [7.1°] and from 83.0° [5.8°] to 81.0° [7.0°], respectively), and piriform width (from 24.5 [1.6] mm to 25.5 [1.3] mm). No significant difference was noted in mean piriform height. We also noted a statistically significant difference between the male and female patients regarding their mean glabellar angle, maxillary angle, and piriform height at both initial and final scans (Table 2). Initial mean (SD) glabellar angle for men was 61.7° (5.7°) compared with 72.7° (5.4°) for women, with final values of 57.9° (4.9°) vs 71.2° (6.0°). Mean (SD) maxillary angle initial values were 87.8° (6.1°) (right) and 87.1° (4.9°) (left) for men, with 79.6° (4.3°) and 80.6° (5.0°) for women, respectively. Final values were 87.0° (4.4°) and 86.9° (4.1°) for men and 77.7° (6.1°) and 77.7° (6.2°) for women, respectively. Mean (SD) piriform height for men was found to be 35.0 (2.0) mm initially and 35.5 (2.1) mm finally, compared with 31.3 (2.8) and 31.6 (3.0) mm for women, respectively. No significant difference between sexes was noted for piriform width.
Correlational statistics were performed comparing our 3-D maxillary angle measurements and our 2-D maxillary angle measurements for the patients for whom 3-D reformatting was obtained. Correlation coefficients of 0.55 and 0.69 for the initial and final scans, respectively, were obtained. We also compared the glabellar angle measurements between the 2-D and 3-D methods and noted a correlation coefficient of 0.90 and 0.91 for the pre- and postscans, respectively.
Although the bony facial aging changes that we define in our study have been previously documented by several groups,1-5 previous skeletal studies have had limiting methodologies, specifically involving CT technology, that resulted in statistically invalid findings. In addition, they were not longitudinal in design because none analyzed the same patient over time. Our pilot study results have confirmed previous findings, as well as documenting a methodology to longitudinally study skeletal changes. As imaging technology continues to advance, we will add other data points as markers to this methodology.
Specific to this study, our 3-D analysis strategy confirmed the decrease in maxillary angle and piriform angle, representing the maxillary retrusion that occurs with age. Because the underlying maxilla provides the essential projection around the nose, these bony changes likely contribute to the appearance of many common mid-face aging changes, such as prominent nasolabial folds,5 facial hollowing, loss of dentition, and the senile nose. These findings are consistent with the documentation by Pessa1 of loss of projection of the maxilla and the quantified description by Mendelson et al4 of maxillary retrusion.
Our 2-D results also showed significant changes between the initial and final scans. We noted a decrease in glabellar and maxillary angles, along with a widening of the piriform aperture. This is consistent with the documentation by Shaw and Kahn2 of these changes in different age groups. We found that the 2-D analysis allowed more precise data gathering than our 3-D analysis because the measuring techniques were more easily standardized. As expected from previous studies with sex comparisons, we did also note a statistically significant difference between the male and female patients.2,4
Interestingly, the strong correlation between the 3-D and 2-D results for glabellar and maxillary angles supports the future use of 2-D measurements as a general marker. Choosing specific points on 3-D reconstructions inherently introduces more variability, and therefore obtaining reliably reproducible measurements was substantially more difficult than on the 2-D films.
Our study was limited by the small sample of patients available in the radiology report database whose records met criteria. This is due to the scarcity of patients who require high-fidelity CT imaging over several years without substantial operative intervention or facial trauma. However, our analytical techniques highlight the possibilities that a multi-institutional database of patients will offer. Our study focused on differences within a patient spanning approximately a decade. Ideally, patients should be studied longitudinally to document changes in the individual. Because aging changes occur over decades, defining methodology using 2-D measurements and adding a larger database would allow a greater range of ages to be included in new studies because older CT images could then meet criteria for inclusion. The fact that our study confirms the results of previous studies that have compared ages ranging from 20 to 70 years is encouraging to further study of specific patterns and timing of facial skeletal changes. In addition, the influence of sex and other demographic characteristics and hormonal and disease states, as well as other variables, can be assessed across individuals.
Ultimately, defining a methodology to longitudinally document the 3-D patterns and timing of facial skeletal aging changes will allow us to objectively test specific treatments aimed at slowing or reversing these bony aging changes. The timing of when intervention can be helpful can also be defined. Treatments already in use for osteoporosis such as hormone modulators, bisphosphonates, or calcitonin may be effective for aging. Mechanical devices used in orthopedics and orthodontics, as well as novel pharmaceutical approaches, may allow us not only to treat but also to prevent these facial skeletal changes from occurring, opening up a whole new paradigm in facial aging prevention.
Corresponding Author: Boris Paskhover, Department of Otolaryngology–Head and Neck Surgery, Rutgers New Jersey Medical School, 90 Bergen St, Suite 8100, Newark, NJ 07103 (firstname.lastname@example.org).
Accepted for Publication: March 29, 2017.
Published Online: August 10, 2017. doi:10.1001/jamafacial.2017.0743
Author Contributions: Dr Paskhover had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Paskhover, Durand, Gordon.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Paskhover, Kamen, Gordon.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Paskhover.
Administrative, technical, or material support: All authors.
Supervision: Paskhover, Gordon.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: