Quantitative computed tomography (QCT) was completed in 34 subjects between the ages of 9 and 33 years with symmetrical mandibles in order to investigate the three-dimensional cortical bone mineral density (BMD) distribution in the mandible. The number and distribution of the pixels were determined at three levels: (1) representing the entire mandibular bone; (2) the cortical bone at 60% above the baseline defined as the segmentation level (around 1050 mg/cm3) and representative of only cortical bone; and (3) the highest mineralized cortical bone (>1250 mg/cm3). The geometrical distribution of the highest mineralized areas was evaluated by three-dimensional reconstruction of the images. The total number of pixels for the entire mandible increased significantly at each time point represented at four increasing ages groups (9–11 years of age, 12–14 years of age, 15–17 years of age, and >18 years of age). The male and female subjects had a similar total number of pixels for the entire mandible before the age of 11, but the male subjects showed a significantly larger total number of mandibular pixels after that age. Comparison of the number of pixels for pure cortical bone (60% segmentation level) and the highest mineralized cortical bone indicated a significant increase with maturation with the greatest change occurring between the 13-year and 16-year age groups. However, the ratio of cortical bone/total bone increased at a more rapid rate in the male subjects and reached a plateau by the 16-year age group, showing distinct differences in mineralization of the mandible between the sexes.
The growth of the mandible has been assessed by various methods that have relied on analysis of the external morphology.(1,2) Considerable attention has been paid to the description of the changes of skeletal size or shape based on two-dimensional studies using full-head X-rays (i.e., cepha-lograms). Two-dimensional radiographic methods are limited in investigating three-dimensional changes that accurately reflect the true growth. Few studies have evaluated bone remodeling beyond changes in its border and shape. Comprehensive evaluation of bone relies on several properties that include assessing its architecture as well as its mineralization.(3–6) More recent techniques with computed tomography (CT) have provided a method to quantify the bone mineralization defined by quantitative computed tomography (QCT) and to evaluate this in a three-dimensional reconstruction.(7–9) Computer programs provide an accurate method to reconstruct the entire mandible from the computer-based images. This three-dimensional reconstruction has extended to developing finite element models (FEM) that predict stresses within the mandible based on computer models that include muscle function.(8,10,11)
Evaluating bone mineralization of the mandible has important clinical implications. Moving teeth orthodontically depends on the level of bone mineralization within the alveolar bone. The potential to fracture the jaw and the rate of fracture healing within the mandible depends on the level of mineralization. The potential to place successfully a metal implant for a prosthodontics device and to predict its acceptance by the mandibular bone can depend on the level of mineralization. The level of mineralization can change with age and, therefore, needs to be evaluated in the growing human and has not been done for the mandible. In our previous study, we introduced a new calibration phantom for applying QCT to the human craniomandibular skeleton.(8) With this technique established, we are now presenting a cross-sectional study as to how the mandible alters its distribution of mineral density with maturation.
MATERIALS AND METHODS
Sixteen female and 18 male subjects between the ages of 9 and 33 years were studied with CT scanning. The children were separated by gender into four age groups: 9–11 years, 12–14 years, 15–17 years, and 18–33 years. The data were obtained and stored between 1990 and 1996 at Showa University Medical Hospital and the Dental School. All subjects had normal craniomandibular growth without medications affecting bone metabolism but had been referred to the hospital for traumatic injury. Their medical history included information related to injuries to the head (e.g., blow to the posterior cranium in a traffic accident) and severe impaction of a tooth. Informed consent for participation of each subject was obtained from the patient and/or parent/guardian. The study received approval from Showa University Ethics Committee and conformed to the standards in the Declaration of Helsinki.
We applied a method that was introduced in our previous study for obtaining images with a calibrated phantom and developing a three-dimensional reconstruction of the mandible to determine the distribution of different levels of bone mineral density (BMD).(8) This method includes a reference phantom specifically designed for the craniofacial region. It has five rods containing calcium hydroxyapatite (HA) with different percentages as determined by weight: 112.8 mg/cm3 (10%), 174.9 mg/cm3 (15%), 241.3 mg/cm3 (20%), 312.4 mg/cm3 (25%), and 388.8 mg/cm3 (30%), respectively. The calibration phantom was positioned next to the head of each subject so both were scanned simultaneously. The CT measures from the phantom were completed daily to determine reproducibility. The CT values varied from 3.0 to 4.0% (Fig. 1). Based on the CT measures (Hounsfield units [HU]) of five points from the phantom cylinders, a linear regression curve was determined between mineralization in milligrams per cubic centimeter of calcium HA and CT HU with a high correlation coefficient (r = 0.99). The effect of changing the position of the reference phantom to the head was evaluated in tests with a circular vessel filled with water surrounded by different positions of the reference phantom cylinders. The correlation coefficient remained high (r = 0.99) despite changing the positions of the cylinders.
The scanning site was verified by first obtaining a full-head lateral scan (e.g., scout or scannogram) that was parallel to the Frankfort-horizontal (FH) plane (upper auditory meatus border to lower border of the infraorbital ridge). Contiguous 2.0-mm-thick slices were obtained through the entire mandibular skeleton. Scanning parameters were set at 120 kilovolt peaks (kVp), 60 mA, and 3.0 s with a 180-mm field of view (FOV) and 512 × 512 pixels (Hitachi, W-600 scanner, Tokyo, Japan and Yokogawa, Quantex scanner, Tokyo, Japan). Original data were stored on magnetic tape. All data including run-length was transmitted to a personal computer system (IBM PC; IBM Corp., Danbury, CT, U.S.A.) for transforming the data into a format for off-line analysis. The transformed data was transmitted to a Macintosh personal computer (PC; Macintosh 8500, Apple Computer, Inc., Cupertino, CA, U.S.A.) for further off-line analysis using custom-designed software (Voxel-View, Vital Images, Fairfield, IA, U.S.A.). The PC system was used to determine the mineral density. The attenuation values from the five-step phantom were transformed into HA values (mg/cm3), and then each pixel attenuation value from the mandible was interpreted as mineral density.
Segmentation of pixels and age-related measurement
All pixels occupying the mandibular bone were selected from the axial images. The threshold value for the automated calculation of the pixels within the cortical bone was determined by applying a linear measure in a given axial image of the lower mandible (Fig. 2A). The minimum level between cortical bone and trabecular bone was selected on a histogram (Fig. 2B, point A). The value at 60% above the baseline of the cortical bone was calculated and this mineralization number was defined as only cortical bone (Fig. 2B, point B). The third value was a high mineralization number (1250 mg/cm3) that correlated to the midrange of the highest calcium concentrations determined by absorption spectrophotometry taken from bone samples of the mandible (Fig. 2B, point C).(8) This procedure was performed routinely in each subject to provide visual images of the mandible at the three levels of mineralization (Fig. 3). The range of pixels for each subject's mandible was displayed as a line graph and the total number of pixels for each mandible was determined for the four age groups for both females and males.
Three-dimensional distribution of high-density area
The cortical bone pixels with values exceeding 1250 mg/cm3 were used to determine the highest mineral density regions of the mandible. These high mineral density pixels were determined within a three-dimensional reconstruction of the entire mandible. The individual horizontal images were reconstructed into a three-dimensional mandible using computer software (Fig. 3). The three-dimensional program allowed increased resolution of selected levels of the mandible based on its mineralization level (Fig. 4).
The total number of pixels for the mandible was displayed as mineralization levels ranging from around 700 to 1700 mg/cm3 with the range varying for each subject. The mean and standard deviation of the total number of pixels for the mandible, cortical bone as defined at the 60% segmentation value, and the highest mineral density cortical bone at and exceeding 1250 mg/cm3 were calculated for each of the four groups of subjects and separated by gender. Comparison among the four age groups first by gender and then between genders was completed with a factorial analysis of variance (ANOVA) (Stat-View, Version 5.0, Abacus Concepts, Inc., Berkeley, CA, U.S.A.). The level of significance was set at p < 0.05 to eliminate type 1 error and the post hoc Fisher test was used to determine the level of significance among the four groups with gender as a covariate.
Number of total bone pixels
The pixels with mineralization values above the segmentation value of around 700 mg/cm3 of HA defined the mandible and included both cortical and some trabecular bone (Fig. 5). Displaying the entire number of pixels over the mineralization level for each subject indicated that this number increased as subjects became older. The highest number of pixels occurred between 700 mg/cm3 and 1100 mg/cm3 of HA with a steady decline after that level. Above 1400 mg/cm3 of HA, all subjects showed a minimal number of pixels that extended to 1700 mg/cm3 of HA. The mean number of total pixels from 770 to 1700 mg/cm3 of HA was determined for grouped data at each age and differentiated by gender (Table 1). The mean total pixel number for the entire mandible significantly increased at each time point (p < 0.0001, Fisher post hoc analysis) for both the male and the female. The greatest increase occurred between the 13-year and 16-year groups. Comparing the female with male subjects showed similar curves of increase but with the female subjects showing a significantly smaller number of total mandibular pixel numbers after the age of 10 years (Fig. 6).
Number of cortical pixels above 60% segmentation value
The pixels within cortical bone at 60% above the threshold mineralization value (around 1090 mg/cm3 of HA) depicted only cortical bone. Comparison of the mean number of pixels for cortical bone at the 60% segmentation level indicated a significant increase at each time point (p < 0.01). The greatest increase occurred between the 13-year and 16-year groups. Comparing the female with male subjects showed similar curves of increase but with the female subjects showing a significantly smaller number of pixels above the 60% segmentation level (Fig. 6; p < 0.05).
Number of high-density pixels in cortical bone
The pixels with mineralization values exceeding 1250 mg/cm3 of HA were compared among the four age groups for both the male and the female. The mean number of pixels for the highest mineralized cortical bone increased significantly at each of the four age groups in the male and in three of the four comparisons in the female (p < 0.01). As with the changes in total pixel number for the mandible and total number of cortical bone pixels exceeding the 60% segmentation value, the greatest increase occurred between the 13-year and the 16-year groups. The distribution of these highest mineralization cortical bone regions also changed with the age groups (Fig. 6).
Change in ratio of number of pixels of cortical bone at the 60% segmentation level and total mandibular pixel number
Comparison among the pixel numbers for the three types of measures indicated that the ratio of number of pixels at the 60% segmentation level/number of total mandibular bone pixels (C/T ratio) showed two distinct patterns between the male and the female over the four age groups (Fig. 7). The female ratio showed a steady increase from the 10-year-old group to the > 18-year-old group with significant differences between groups. In contrast, the males showed a steeper change among the first three age groups that were significantly different and then no further change with a plateau in the ratio by the 15-year and > 18-year groups.
Change in ratio of number of pixels with the highest density cortical bone and total mandibular pixel number
The ratio of the number of high-density cortical bone pixels/number of pixels at the 60% segmentation level (H/C ratio) showed two different patterns of change over the four age groups between the male and the female. The female ratio remained relatively constant over three age groups and only significantly increased with the females 18-year and older group (Fig. 7). In contrast, the male ratio stayed the same for the two youngest groups of 10-year olds and 13-year olds, and then significantly increased starting with the 16-year-old group and increasing further in the >18-year-old group.
Three-dimensional distribution of high-density area
The three-dimensional distribution of highest mineralized cortical bone only included certain regions of the mandible (Fig. 8). Subjects at all ages showed the highest density of cortical bone within the symphysis and body of the mandible and along the lower anterior border of the ramus. Beyond these regions the presence of the highest density cortical bone varied with age and gender.
Mandibular growth analysis
The mandible or jaw is a separate bone of the head that articulates on two bilaterally situated joints and contains two arches of teeth that interdigitate.(12) The mandible is a curved long bone with no osseous articulations inferior to the plane of occlusal loading.(13) Bending and torsional loads generate the highest stresses. More than 20 muscles insert in and around the mandible with some providing forces that load reaction forces on the dentition and the condyles of the temporomandibular joint. These muscles also transmit direct forces to the bone as measured by surface strain gauges.(14) The mandible can develop fully in a fetus with minimal developing muscle tissue, indicative of the genetic control for expression of this bone.(15) However, without functioning muscle tissue, the mandible modifies its shape and thickness indicating the significance of the muscles to the normal developing jaw.(16) Supporting evidence of muscle modifying the bone's shape and geometry have evolved from computer models of other bones in which the computer-aided optimization to predict femoral geometry during growth is unable to simulate the morphology of the femoral shaft without the addition of muscle forces.(17)
Various methods have been employed to evaluate mandibular growth and its change in shape and size.(2) From histochemical studies, there are two types of bone growth in the mandible.(18) One is appositional and resorptive remodeling at the periosteal and endosteal surfaces. The second is cartilaginous growth in the condyle. In the most common clinical evaluation using full-head X-rays, changes in the size and shape are evaluated by determining the external bone growth.(2,19,20) Superpositioning of X-ray tracings taken periodically over various bony landmarks is useful to obtain the growth pattern of the mandible. This method allows a description of the increase in size and the change in proportion of the same growing bone or group of bones in the craniomandibular complex. It reveals the rate, the amount, and relative direction of growth. However, such routine head films assessed in clinical observation are two-dimensional, and this is a limitation. Only the anterior and posterior edges of the mandible can be visualized in lateral cephalograms. The cortical bone between these edges cannot be visualized and their internal remodeling and/or changes in mechanical quality cannot be determined. In contrast, applying QCT to evaluate mineralization within a three-dimensional reconstruction provides a method to assess the mandible.(7–9,21,22)
To fully understand bone remodeling, the ideal methods should include structural information (i.e., size, shapes, architecture, and connectivity) and characteristic properties (i.e., density, hardness, and elasticity) that are applied in a three-dimensional reconstruction. We attempted this approach in our first paper evaluating QCT with a specialized phantom.(8) Although this method has many advantages, it has limitations including beam hardening and partial volume averaging that are related to the pixel size. The determination of cortical bone density depends on the thickness of the sample and adjusting for partial volume measures that include more than cortical bone. The pixel size used in our study was 0.35 mm × 0.35 mm, providing a high resolution and less chance of partial volume effects. Our method for segmentation of the cortical and trabecular bone at a 60% segmentation level, as described by Louis et al., provided a reliable and logical method to account for partial volume measures.(23)
Mineralization of the growing mandible
Although previous studies have emphasized the gross changes in the external shape of the mandible related to external forces, a second major indicator of the mandibular bone is its mineralization. Mineralization of bone depends on the matrix organization.(13) The slower the appositional rate of bone, the more highly organized the matrix and the greater the strength of the bone. Slow forming of bone tissue leads to development of the high-quality, lamellar type. Lamellar bone achieves greater mineral density and stiffness than less organized forms of woven bone. Primary mineralization refers to the initial 70% of bone mineral deposited at the time of bone formation. The secondary mineralization completes ossification of the matrix and occurs by 8-12 months. Mineralization is sensitive to forces and loads.(5,21,24–29) Studies of other skeleton bones assessing mineralization indicate that exercise and use of the bone effect mineralization.
The relevance of differences in bone mineralization to functional loads has been studied in a cause-effect relationship in few studies. One exception was the evaluation of a female patient who experienced a knee ligament rupture during a long-term strength-training program.(27) Evaluation of muscle force and BMD before and after unloading and recovery showed two important points. First, bone mineral apparent density decreased by around 25% over 3-months after the initial injury, which prevented any use of the limb. By 4-months, the patient reinstituted strength training, and by 11 months, the muscles had achieved full strength. BMD remained below its original control levels (around 12%) and did not fully recover until 2 years after the injury. This study suggests that bone mineralization is a long-term adaptation that follows changes in muscle strength but after a considerable lag period. A long delay occurs between the full strength of the muscle to generate force on its bone and the development of a fully mature and mineralized bone.
Table Table 1.. Number of Pixels (Mean ± SD)
> 18 years
Group age (year) (n)
1: All comparisons significant P < 0.01 to <0.0001. 2: 15-17 versus >18, NS; all other comparisons significant. 3: 9-11 versus 12-14, NS; all other comparisons significant. 4: Only 15-17 versus >18 significant.
Number of pixels
Total bone pixels (T)
54,350 ± 6898
77,411 ± 9589
118,402 ± 10,643
132,484 ± 5840
Cortical bone pixels (C)
14,681 ± 2916
25,426 ± 4591
44,033 ± 5476
48,081 ± 2623
High density pixels (H)
2810 ± 802
4831 ± 1181
9500 ± 1078
11,909 ± 1378
0.27 ± 0.02
0.33 ± 0.03
0.37 ± 0.02
0.36 ± 0.02
High density/cortical (H/C)
0.19 ± 0.02
0.19 ± 0.02
0.22 ± 0.02
0.25 ± 0.03
Number of pixels
Total bone pixels (T)
43,796 ± 4644
57,720 ± 6990
96,492 ± 3591
109,684 ± 3703
Cortical bone pixels (C)
11,863 ± 1945
16,530 ± 3089
29,832 ± 2089
36,491 ± 1903
High density pixels (H)
2306 ± 497
3321 ± 616
5930 ± 841
8194 ± 1023
0.27 ± 0.02
0.29 ± 0.04
0.31 ± 0.02
0.33 ± 0.01
High density/cortical (H/C)
0.19 ± 0.02
0.20 ± 0.02
0.20 ± 0.02
0.23 ± 0.03
The effect of mechanical loading of bone by muscle depends on a systemic and metabolic background controlled by hormones as proposed in the mechanostat theory of bone.(30,31) Bones vary regionally in their susceptibility to control by hormonal and mechanical loading.(32) This finding suggests that evaluating regional differences in bone mineralization of the mandible will provide insight into the history of different types of loads and stresses that are continually applied to the mandible over a long period of time against the underlying hormonal control. Our data provides the first detailed evaluation of mineralization across the entire mandible and in subjects that are young during the rapidly growing ages. Our work shows that cortical bone develops differently in the female versus the male. This may suggest hormonal factors such as estrogen that are effective in modulating muscle forces during development.(33,34)
As Fig. 8 shows, the distribution of the highest mineralized cortical bone within the mandible will vary among individuals. However, the highest mineralized cortical bone always occupies certain regions of the mandible that include the corpus of the mandible and the anterior border of the ramus. This pattern extends to include more regions of the mandible during growth. Presently, we are determining if particular patterns of distribution are peculiar to females versus males. Earlier studies have assessed mineralization of the mandible by CT and similar techniques but only in adult humans, only in selected sites, and not with three-dimensional reconstruction.(35) The value of applying QCT in a three-dimensional reconstruction is the ability to evaluate overall growth in more than two planes and to evaluate the actual growth pattern while assessing regional changes in mineralization. In this article our assessment of increases in pixel number over age is the first stage in evaluating the overall multidirectional growth of the mandible. Our emphasis on how cortical mineralization changes with this growth is to evaluate how the craniofacial region alters the strength of bone during maturation. The QCT data also can be applied to stress analysis within the craniofacial region and is an area of our active research. Our results confirm that the growing subject attains a high proportion of adult bone density by his and her late teenage years. Nine to eleven-year-old males and females have approximately 33–35% of the adult cortical bone volume. By 15–17 years of age, they have achieved approximately 83–87%.
Our results give the most accurate indication of the development of the mandible based on its total structure represented by multiple pixels defined in a three-dimensional reconstruction. The most rapid change in the mandibular size, including its cortical bone, occurs between 11 and 16 years for both the male and the female. Interestingly, the males develop more of the highest mineralized regions sooner than the females beginning at 11 years whereas the female starts after 16 years of age. We suspect that the males increase the cross-sectional area and total volume of their jaw-closing muscles (as has been documented in animal studies) greater than in the female with the resulting effect of developing greater loads and stresses within the mandible. Presently, we are evaluating the cross-sectional area of these normal subjects. When BMD, in terms of area BMD, (g/cm2) was evaluated for the lumbar spine and femoral neck using dual-energy X-ray absorptiometry (DEXA) in 65 normal children between the ages of 7 and 20 years, the females showed the most marked annual increase between 11 and 13 years of age at the time of menarche.(36) In contrast, the males showed their largest increases slightly later during the 13–17 years of age. The distinction in development of mineralization in the mandible compared with the lumbar spine and femoral neck between the two sexes may suggest the relevance of muscle function to mineralization of the three respective regions. This concept is supported by a regression analysis of factors associated with mineralization in the spine, greater trochanter, femoral neck, and radius of 90 children between the ages of 6 and 14 years. The study showed varying effects on skeletal mineralization depending on the skeletal site.(26) Increases in calf muscle area were strongly associated with mineralization. Physical activity and puberty effected mineralization varying with the four sites, indicating the complexity of muscle activity on mineralization.
Three-dimensional distribution of high mineral density regions
Measurement of bite forces, measurement of bone surface strain using strain gauges, and computer models using static equilibrium theory provide indirect assessment of forces applied to the craniomandibular skeleton. Present knowledge suggests that different regions of the craniomandibular skeleton exhibit different propensities to forces modifying the bone.(14) Research in monkeys indicates that a region like the supraorbital ridge exhibits little strain during a high force-developing response like chewing, suggesting that its mass is primarily determined genetically. Surface strain gauge studies indicate that the mandible can flex, twist, compress, and stretch in different regions during common functions such as chewing. Lighter strains develop during swallowing and speech. Surface strain gauge measures over the symphysis, along the mandibular cortical bone beneath the first molar, and in the neck under the condyle indicate that the same region of bone can undergo tensile, compressive, and shear strains at different times in even one function such as chewing. The actual relevant biological signals involved with translating the physical changes into cellular and molecular signals that mobilize osteoblasts and osteoclasts as well as their precursors have been suggested but are under present investigation. Such studies suggest some interaction between forces developed within bone and its shape and size, but we feel that a more accurate indicator of how forces modify bone may rely on the distribution of mineralization within the cortical bone of the mandible, and that this individual pattern of distribution in each subject will provide a highly accurate indication of the impact of muscle forces effecting the bone.
Only a few studies have evaluated mandibular bone density in living subjects. Most of these studies have used dual proton absorptiometry (DPA) or DEXA.(35,37,38) These methods are only two-dimensional evaluations, limited in accurate measurement, of a complex geometry of the craniomandibular skeleton. QCT has the advantage that it can be used to develop an accurate three-dimensional reconstruction and also can be applied to the development of an FEM that allows the evaluation of forces and stresses developed within the bone. Biomechanical studies have analyzed the controlling factors of secondary osteon bone clarifying the relation between mechanical strain and remodeling bone and mechanical properties (e.g., Young's modulus). The application of mathematical methods such as the FEM has improved our interpretation of bone remodeling and such models use the mechanical properties in the analysis.(10,39) However, only a few methods have been able to use the most accurate approach of each patient's individually reconstructed CT data for the FEM. We have been working on this approach, and our next publication will discuss the application of our CT data to an FEM relevant to each subject.
Our initial evaluation of the distribution of the highest mineralization sites indicates that it changes with development of the mandible. It does not seem to correlate with muscle insertion points except possibly along the anterior ramus where the temporalis muscle attaches, but further analysis of this pattern over age is being developed in our next publication. Previous evaluation of the growing mandible from gross anatomical structure has led Enlow to express the opinion that the sites of bone apposition and resorption in the mandible do not relate to the insertion or origin of the jaw-closing muscles.(2) The muscles attaching to the mandible have effects well beyond their attachment site and develop the variety of strains measured in the monkey. Those strains suggest different loading and stresses throughout the mandible.
This experimental cross-sectional study shows that when CT is combined with a phantom of hydroxyapatite, the pixels that represent bone within the human mandible can be defined in terms of mineralization. The range of mineralization increased with age. The mean total pixel number for the entire mandible significantly increased at each time point with the greatest increase occurring between the 13-year and 16-year groups. Comparing the female with the male subjects showed similar curves of increase but with the female subjects showing a significantly smaller number of total mandibular pixels after the age of 10 years. The ratio of the number of high-density cortical bone pixels/total mandibular bone pixels showed two different patterns of change over the four age groups between the male and the female. The female ratio remained relatively constant over three age groups and only significantly increased with the females' 18 years and older group. In contrast, the male ratio stayed the same for the two youngest groups of 10 and 13 year olds and then significantly increased starting with the 16-year-old group and increasing further in the >18-year group. The male subjects showed an earlier age at which the highest mineralized cortical bone developed than the females.
This work was supported by the Japanese Ministry of Education and was conducted in collaboration with the Departments of Orthodontics at Showa University, the Mechanical Engineering Department at the Tokyo Institute of Technology, and the Department of Orthodontics at the University of California at San Francisco. This study was supported by the Mnistry of Education, Grant-in-Aid for General Scientific Research (01440082).