Key words:
Bone density - cone-beam computed tomography - dental implants - predictability
INTRODUCTION
Since the introduction of osseointegration field by Branemark 1987, Restoring missing
teeth with dental implants has become a predictable treatment option,[1] by which function and esthetics can be restored successfully. From a clinical point
of view, successful osseointegration is reflected by implant stability, which can
be divided into primary and secondary stability. Primary stability is associated with
the mechanical engagement of an implant with the surrounding bone by which micromotion
is reduced, allowing for the biological process of osseointegration of regeneration
and remodeling to achieve the secondary (biological) stability.[2]
Several factors have been found to affect the primary stability of endosseous implants
which can be classified into patient-related (like bone quality and quantity) and
non-patient related factors (like implant design and surgical procedure) . Regarding
patient-related factors, available bone quality and quantity are considered prerequisite
for successful and predictable implant treatment.[3] While the available bone quantity is reflected by the dimensions of alveolar bone
that will house the fixture part of the implant, bone quality has been described by
the relative amount of cancellous and cortical portions and the density of the recipient
site of the alveolar bone.
In addition to the positive association between bone density and implant stability,[4]
[5]
[6]
[7] bone density can affect implant healing time and can modify the surgical technique.[8] Moreover, bone density evaluation in the periapical area of teeth has its importance
in endodontics,[9] and bone density evaluation at specific areas like the palate, can be helpful in
locating placement sites for mini implants and orthodontic skeletal anchorage devices.[10]
For objective assessment of bone density at implant sites, both computed tomography
(CT) and cone-beam CT (CBCT) can be used, and good correlation was shown between CT
numbers (Hounsfield unit [HU]) and CBCT gray values.[11] Unlike CT, many factors can affect CBCT grey or intensity values like type of scanner,
field of view, position, and scanning parameters.[12] Calibration is needed to generate CBCT intensity values similar to CT HUs.[13]
On the other hand, several classifications have been proposed for subjective bone
density assessment.[14]
[15]
[16]
[17] In the recent study done by Rebaudi et al.,[17] a novel method of classifying bone density in CT/micro-CT into hard, normal and
soft was proposed and suggested to be used in combination with objective bone density
analysis. Even in case of CBCT, objective assessment of the bone density should be
added to subjective assessment due to limitations in subjective assessment.[18]
[19]
[20]
Similar to the example of Rebaudi et al.,[17] we classified bone density at implant sites into low, intermediate, and high in
the current study. To the best of our knowledge, no study has investigated the predictability
of these three categories of bone density using CBCT intensity values, thus validating
the conduction of this study.
MATERIALS AND METHODS
Patients
In our retrospective study, images for all patients who underwent CBCT examination
for dental implant treatment and other dental purposes at our dental radiology clinic
were retrieved and evaluated between January 2011 and January 2016. Only cases with
missing lower posterior teeth were included in the study. The number of cases included
were 436 (160 premolars and 276 molars), from 210 patients (85 males and 125 females)
with a mean patient age of 46 years. Cases with artifacts or pathology affecting bone
density at implant sites were excluded from the study. In addition, we excluded cases
in which we were not able to simulate placement of 4 mm × 10 mm dental implant. This
study was approved by our institutional research board (no. 382/2016).
Cone-beam computed tomography examination
As a CBCT apparatus, KODAK 9500 Cone Beam 3D System (Carestream, Rochester, NY, USA)
with flat panel detector was used. The imaging area of CBCT is a cylinder with a height
of 15–20.6 cm and a diameter of 9–18 cm providing isotropic cubic voxels with sides
approximating 0.2–0.3 mm. Only cases examined with 0.2 mm were included in the study.
The exposure parameters were: 90 kV as a tube voltage, 10 mA as a tube current, and
10.8 s as an exposure time.
Examinations were performed by 360° rotation in the occlusal position with the patient
standing and closing their teeth.
Images
One calibrated oral radiologist (MA) with 9 years of experience with CBCT and dental
implants was responsible for determining the implant sites, subjectively classifying
the bone density on cross-sectional images at proposed implant sites into: Low, intermediate,
and high, generating CBCT intensity values based on this classification, then saving
the images for a second evaluation after 1 month, and third evaluation by one calibrated
oral implantologist (MH) with 9 years of experience on a separate occasion. When the
two observers disagreed about bone density evaluation, they evaluated the images again,
and a consensus was then reached by discussion.
For the subjective evaluation of bone density at the implant sites, the trabecular
bone was considered low density if marrow spaces are filling most of the site, intermediate
density if bone trabeculae are filling half of the site and high density if bone trabeculae
are filling most of the site [Figures 1]
[2]
[3].
Figure 1: Cone-beam computed tomography section showing high dense trabecular bone at mandibular molar implant site
Figure 2: Cone-beam computed tomography section showing intermediate dense trabecular bone at mandibular premolar implant site
Figure 3: Cone-beam computed tomography section showing low dense trabecular bone at mandibular molar implant site
Using InVivo software (Anatomage, San Jose, California, USA), arch section module
was utilized for determining the implant sites, simulating implant placement and generating
CBCT intensity
values. For simulating implant placement; the distance measurement tool was used for
drawing a 4 mm × 10 mm rectangle at the implant sites, then HU measurement tool was
used to generate intensity values for the simulated implants. HU measurement tool
showed 3 intensity values at each implant site (minimum, mean and maximum). However,
we only considered the mean intensity value for analysis.
All images were evaluated on high definition liquid crystal display with installed
Invivo software, and window settings were fixed for all cases.
Statistical analysis
Analysis of data was achieved through the Statistical Package for Social Sciences
software (version 15; SPSS Inc., Chicago, IL, USA). Means and standard deviations
(SDs) as well as percentages were used to describe data. The difference in intensity
values between the low, intermediate, and high- density sites was analyzed using one-way
ANOVA. Overall percent agreement and Kappa statistics were used to determine the measure
of agreements in the subjective evaluation of bone density between the two observers.
Receiver-operating characteristic (ROC) curve analyses were used to examine the overall
predictive power, sensitivity and specificity, and corresponding cutoff points of
CBCT intensity values. The overall performance of CBCT intensity values for predicting
bone density was assessed by computing the area under the curve (AUC). The best cutoff
points for CBCT intensity values were determined at the point on the curve where the
sum of sensitivity and specificity was highest. A value of P < 0.05 was considered statistically significant.
RESULTS
The overall percent of agreement in subjective classification of bone density for
the first observer over the two repeated occasions of evaluations was 66.5% [Table 1] with a moderate agreement between the two repeated evaluations (Kappa statistics
= 0.50; P < 0.005). The overall percent of agreement between the two observers was 64.7% [Table 2] with a moderate agreement between the two observers (Kappa statistics = 0.47; P < 0.005).
Table 1:
The overall agreement for the first observer between the two repeated occasions of
measurements
|
Density
|
Second time
|
|
Low
|
Intermediate
|
High
|
|
n
|
%
|
n
|
%
|
n
|
%
|
|
First time
|
|
Low
|
67
|
15.4
|
88
|
20.2
|
0
|
0.0
|
|
Intermediate
|
4
|
0.9
|
110
|
25.2
|
40
|
9.2
|
|
High
|
0
|
0.0
|
14
|
3.2
|
113
|
25.9
|
Table 2:
The overall agreement between the two observers
|
Density
|
Second observer
|
|
Low
|
Intermediate
|
High
|
|
n
|
%
|
n
|
%
|
n
|
%
|
|
First observer
|
|
Low
|
64
|
14.7
|
89
|
20.4
|
2
|
0.5
|
|
Intermediate
|
8
|
1.8
|
104
|
23.9
|
42
|
9.6
|
|
High
|
0
|
0.0
|
13
|
3.0
|
114
|
26.1
|
Based on the consensus of the two observers, 15.6% of sites were of low bone density,
47.9% were of intermediate density, and 36.5% were of high density. The means (SD)
of intensity values were 172 (99.3) for those with low density, 307 (123.1) for intermediate
density, and 645 (192.3) for high density (P < 0.05).
ROC analysis showed that CBCT intensity values had a high predictive power for predicting
high density sites (AUC = 0.94, P < 0.005, [Figure 4]) and intermediate density sites (AUC = 0.81, P < 0.005, [Figure 5]). The best cutoff value for intensity to predict intermediate density sites was
218 (sensitivity = 0.77 and specificity = 0.76) and the best cutoff value for intensity
to predict high density sites was 403 (sensitivity = 0.93 and specificity = 0.77).
Figure 4: Receiver-operating characteristic curve for predicting high density implant sites
Figure 5: Receiver-operating characteristic curve for predicting intermediate density implant sites
DISCUSSION
The assessment of bone density at potential dental implant sites is considered of
paramount significance presurgically since it affects locating the best implant site,
implant site preparation technique, implant positions, and the success rate of the
implants.[17]
[21]
[22]
[23] Therefore, several studies have been conducted for this task,[4]
[5]
[6]
[7]
[8]
[17]
[23]
[23]
[20] and different results were revealed due to use of various scanners, softwares, and
methods.
Among imaging modalities used for bone density assessment, CBCT has advantages over
conventional CT due to high image resolution and low radiation dose,[24] and an advantage over micro-CT, since it is being used clinically and not only for
in vitro experiments. Moreover, CBCT showed comparable results to micro-CT in assessing grey
level distribution in human mandible.[25]
In this study, we evaluated bone density at posterior mandibular implant sites. Around
half of the sites had intermediate density, and the remaining half had both of low
and high-density type of bone. This shows the importance of site-specific evaluation
as recommended in previous studies.[17]
[20]
[22]
[26]
The inclusion of only posterior mandibular implant sites allowed us to simulate placement
of one of commonly used implant sizes for all cases, and to evaluate the density of
crestal 10 mm of trabecular bone, which is the most important area for osseointegration.[27] In addition, the alveolar ridge of posterior mandibular implant sites is vertically
oriented, which enabled us to simulate placement of vertically oriented implants and
to overcome the limitation of the software being used, in which it cannot simulate
placement of tilted implants.
For the subjective assessment of bone density in the current study, the intra- and
inter-observer agreement were moderate; this reveals the difficulty in subjective
assessment, especially when the bone has low to intermediate or intermediate to high
density. The difficulty in subjective assessment of bone density was also present
in few previous studies,[8]
[26]
[28]
[29] which could partially be due to dependence of subjective assessment on observer
experience.
To overcome the limitations and difficulties in subjective visual assessment of bone
density, the use of an objective scale like the one suggested by Norton and Gamble,[28] or Trisi and Rao[29] is useful and would be more accurate. However, this cannot be applied to CBCT. Similarly,
we investigated the usefulness of CBCT intensity values in predicting bone density,
and the values had high power in prediction of different types of bone density.
In the previous study by de Oliveira et al.,[26] they used the subjective classification of bone density suggested by Lekholm and
Zarb,[15] and they divided the bone into four types. Despite this, they had only three categories
of HU; below 200 for Type 4, between 200 and 400 for Type 2 and 3, and more than 400
for denser bone (Type 1). Again, similarly to Norton and Gamble[28] and Trisi and Rao,[29] they had difficulty in the subjective differentiation, specifically between Type
2 and 3 of bone. Therefore, our classification of bone density into low, moderate,
and high, would be easier and more flexible to apply than classifying the bone into
four types. Interestingly, our results are in agreement with de Oliveira et al.,[26] since our cutoff intensity values were 403 and 218 for high and intermediate dense
bone, respectively. This might be due to the exclusion of cortical bone from density
assessment. Nevertheless, this shows the usefulness and ability of CBCT to generate
intensity values comparable with CT numbers.
In the present study, a difference of about 200 of CBCT intensity value was required
to differentiate between one type of bone density and the consecutive type. This difference
is close to what reported by Lee et al.,[30] as they reported a difference of 180 of HU.
In contrast to other previous studies,[18]
[20]
[21]
[26]
[28] we simulate placement of implants in the crestal 10 mm of the trabecular bone, without
including the cortical bone. The inclusion of cortical bone will increase the mean
intensity value and is one cause for the different results. Needless to say, the shape
and/or size of region of interest, and the section being used for evaluation are other
causes for different results.
Our study has some limitations which require mentioning. We used third party software
for CBCT intensity values calculation; this affected the quality of the imported images
and resulted in difficulty in subjective image evaluation. Moreover, we simulated
placement of a specific size of implants, which may not necessarily represent a true
clinical situation and other factors may affect the treatment plan. Finally, our results
cannot be generalized unless same CBCT machine and same protocol is followed for obtaining
CBCT intensity values.
CONCLUSION
In conclusion, CBCT intensity values can be used for predicting bone density at posterior
mandibular implant sites.
Acknowledgment
We are thankful for the Deanship of Scientific Research in Jordan University of Science
and Technology for their approval and support in conducting the research.
Financial support and sponsorship
Nil.
