www.nature.

com/scientificreports

OPEN The pterygomaxillary junction:

An imaging study for surgical
information of LeFort I osteotomy
Received: 20 February 2017 Yen-Po Chin1, Maria Belen Leno2, Sarayuth Dumrongwongsiri2, Kyung Hoon Chung2, Hsiu-
Accepted: 11 August 2017 Hsia Lin3 & Lun-Jou Lo2
Published: xx xx xxxx
Maxillary osteotomy is a common surgical procedure and often involves separation of the
pterygomaxillary junction (PMJ), which is a “blinded” procedure with inherent risks. Knowledge of the
PMJ structure is essential. It remains unclear whether patients with different facial types have different
PMJ structures, or different surgical outcome. This study evaluated the computed tomographic images
of 283 consecutive patients who received orthognathic surgery. Patients were classified into Angle
class I, II, III and cleft lip/palate groups. The results showed that the PMJ was 5.1 ± 1.4 mm in thickness,
9.7 ± 1.7 mm in width, and 102.0 ± 4.0 degrees relative to the sagittal plane in the level of posterior
nasal spine. There were no statistically significant differences in these measurements among the
groups. The class III group demonstrated significantly smaller angle relative to the maxillary occlusal
plane. The cleft group showed significantly longer vertical distance between the posterior nasal spine
and the lower border of PMJ, shorter distance between the second molar and PMJ, and longer distance
between the descending palatine artery and PMJ. With regard to postoperative outcome, the cleft
group showed higher incidence of pterygoid plate fracture. The results in this study provide additional
surgical anatomic information.

Maxillary LeFort I osteotomy is a widely applied procedure in the surgical management of dentofacial deform-
ities, trauma, as well as tumors in the skull base and midfacial regions1, 2. Nevertheless, separation of the ptery-
gomaxillary junction (PMJ) during the osteotomy is a challenging procedure for most surgeons, as it sits behind
the maxilla and is not directly visible. While most of the LeFort I related complications could be avoided by an
adequate planning and implementation, the PMJ separation is a blind maneuver and technically risky, especially
in patients with craniofacial deformities.
During the pterygomaxillary osteotomy, the pterygoid plate should remain intact3. A careless attempt to sep-
arate the PMJ could result in untoward pterygoid plate fracture4–7, vascular or neural complications8–14, and even
blindness15–17. The patterns of pterygoid plate fractures have been evaluated in radiographic, cadaveric, and dry
skull studies6, 18–22. However, these studies were mostly made in normal samples, instead of patients with maloc-
clusion or with cleft lip and palate deformities. Likewise, fracture occurring in front of the PMJ causes surgical
difficulty in maxillary movement. Patients with cleft deformity have been proven to be at relatively high risk for
the complications related to atypical anatomy23. Due to the high prevalence of requiring LeFort I osteotomy in
patients with cleft lip and palate deformity, malocclusion, or facial asymmetry24–26, accurate surgical disjunction
of the PMJ can prevent unexpected complications.
Several methods have been described in the literature to maximize patient safety and prevention of complica-
tions since Turvey first described recommendations to safely perform the LeFort I osteotomy27. In the literature,
operative techniques varied with using an oscillating saw, an endoscope approach, or avoiding direct PMJ sepa-
ration by using various techniques28–35. Some techniques were markedly time-consuming, surgeon dependent, or
relied on the availability of specific instruments. However, the results have not been consistent regardless of the
approach36, 37.

1
Department of Medicine, National Yang Ming University, Taipei, Taiwan. 2Plastic & Reconstructive Surgery, and
Craniofacial Research Center, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan. 3Craniofacial
Research Center, Chang Gung Memorial Hospital, Taoyuan, Taiwan. Yen-Po Chin and Maria Belen Leno contributed
equally to this work. Correspondence and requests for materials should be addressed to L.-J.L. (email: lunjoulo@
cgmh.org.tw)

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 1

www.nature.com/scientificreports/

Class I Class II Class III Cleft Total p value AVONA

Demographic variables
Sample (sides) 124 136 168 138 566
Male/Female 17/45 12/56 42/42 39/30 110/173
Mean age (year) 26.1 ± 5.1 26.0 ± 6.2 23.3 ± 5.0 18.0 ± 2.4 23.5 ± 4.8
Anatomic variables
Posterior nasal spine level
Thickness (mm) 5.2 ± 2.0 5.2 ± 0.9 5.1 ± 1.6 4.9 ± 1.3 5.1 ± 1.4 p > 0.05
Width (mm) 9.9 ± 1.6 9.5 ± 1.6 9.7 ± 1.7 9.5 ± 1.7 9.7 ± 1.7 p > 0.05
Angle relative to the sagittal plane (degree) 100.7 ± 4.8 104.0 ± 4.0 102.1 ± 4.3 102.0 ± 4.6 102.0 ± 4.0 p > 0.05
Lower border of the pterygomaxillary junction level
Thickness (mm) 5.9 ± 1.0 6.1 ± 1.3 6.1 ± 1.6 5.9 ± 1.4 6.0 ± 1.4 p > 0.05
Width (mm) 7.8 ± 1.7 8.3 ± 1.5 8.3 ± 1.6 8.3 ± 1.6 8.2 ± 1.6 p > 0.05
Angle relative to the sagittal plane (degree) 117.1 ± 6.9 120.0 ± 6.7 119.3 ± 7.1 115.1 ± 8.6 118.1 ± 7.3 p > 0.05
Other measurements
Distance between P level and L level (mm) 6.3 ± 1.5 6.3 ± 1.4 5.8 ± 1.8 8.4 ± 2.0 6.4 ± 1.9 p < 0.001
Distance to the distal aspect of the second molar (mm) 11.0 ± 1.5 11.6 ± 1.8 11.1 ± 1.6 9.3 ± 2.0 10.9 ± 1.8 p < 0.001
Distance to greater palatine foramen (mm) 2.8 ± 1.9 3.2 ± 0.6 3.0 ± 0.3 3.4 ± 1.0 3.1 ± 0.8 p < 0.001
Angle relative to the maxillary occlusal plane (degrees) 108.7 ± 6.4 93.2 ± 9.2 84.4 ± 2.0 99.6 ± 5.1 95.0 ± 5.5 p < 0.001

Table 1. Demographic and anatomic variables of the pterygomaxillary junction of Angle class I, class II, class
III and cleft patients. All data showed in mean ± SD. P level = posterior nasal spine level. L level = lower border
of the pterygomaxillary junction level.

A thorough understanding of the PMJ anatomy is mandatory in patients requiring LeFort I osteotomy. The
literature remains unclear whether different types of craniomaxillofacial deformity affect the PMJ structure or
intra-operative pterygoid plate fracture. In this study, cone-beam computed tomogram (CBCT) images of 283
consecutive patients undergoing orthognathic surgery were analyzed. This information should provide surgeons
with a better understanding of the PMJ region.

Results
The patient numbers, age, and gender of these groups were listed as following: 62 patients (17 M 45 F, mean age
26.1 ± 5.1 years, range 15 to 38 years) in the Angle classification class I group, 68 patients (12 M 56 F, mean age
26.0 ± 6.2 years, range 15 to 39 years) in the class II group, 84 patients (42 M 42 F, mean age 23.3 ± 5.0 years,
range 17 to 38 years) in the class III group, and 69 patients (39 M 30 F,mean age 18.0 ± 2.4 years, range 15 to 32
years) in the cleft lip/palate group. Demographic and anatomic variables for each group were summarized in
Table 1. For further analysis in the cleft group, we divided the cleft group into the cleft side group, the non-cleft
side group, and the bilateral cleft side group. The demographic and anatomic variables were demonstrated in
Table 2. In general, the PMJ was 5.1 ± 1.4 mm in thickness, 9.7 ± 1.7 mm in width, and 102.0 ± 4.0 degrees rela-
tive to the sagittal plane in the level of posterior nasal spine. There were no statistically significant differences in
these measurements (p > 0.05, ANOVA test) among the Angle class I, class II, class III and the cleft group at the
posterior nasal spine level (P level) and the lower border of the PMJ level (L level) (Table 1). Statistically signifi-
cant differences in the thickness, width and the angle relative to the sagittal plane were noted between the P level
and L level (all p < 0.001, ANOVA test) (Fig. 1a to c).There was statistically significant difference in the class III
group in the angle relative to the maxillary occlusal plane comparing with other three groups (p < 0.001, ANOVA
test) (Fig. 1e).The distance to the distal aspect of the second molar of the cleft group was significantly shorter
(p < 0.001, ANOVA test) (Fig. 1d), the distance to the greater palatine foramen of the cleft group was significantly
longer (p < 0.001, ANOVA test) (Fig. 1g), and the distance between the P level and the L level of the cleft group
was significantly longer (p < 0.001, ANOVA test) compared to the other three groups (Fig. 1f).
For the analysis within the cleft group, the thickness of the cleft side group at the P level was significantly thinner
compared to that of the non-cleft side group and the bilateral cleft side group (p < 0.001 and p < 0.01 each, ANOVA
test) (Fig. 1h). Significant differences were found in the thickness, width and angle relative to the sagittal plane
between the P level and the L level (p < 0.001, p < 0.05 and p < 0.001 respectively, ANOVA test) (Fig. 1h, i and j).
There was no significant difference found in the width and angle within cleft groups (p > 0.05, ANOVA test)
(Fig. 1i). There were also no significant differences found in the distance to the distal aspect of the second molar
and the angle relative to the maxillary occlusal plane in the analysis within cleft group (p > 0.05, ANOVA test)
(Fig. 1j and k).
The postoperative CBCT axial images were used at the P level to study the pterygomaxillary separation type
after LeFort I osteotomy. Postoperative images of the Angle class I group (24 images, 48 sides), class II group (53
images, 106 sides), class III group (84 images, 168 sides) and the cleft lip/palate group (69 images, 138 sides) were
collected (Table 3). The cleft lip/palate group was composed of unilateral cleft lip/palate group (52 images, 52 cleft
sides, 52 non-cleft sides) and bilateral cleft lip/palate group (17 images, 34 bilateral cleft sides) (Table 4). The frac-
tures were classified into three types including clean-cut type, maxillary sinus type, and pterygoid plate type. The
clean-cut type involved the most portions in the class I, class II, class III and the cleft group (60%, 69%, 67% and

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 2

www.nature.com/scientificreports/

Cleft side Non-cleft side Bilateral cleft side p value ANOVA

Demographic variable
Sample(sides) 52 52 34
Male/Female 32/20 32/20 6/11
Mean age (year) 18.0 ± 2.4 18.0 ± 2.4 18.0 ± 3.4
Anatomic variable
Posterior nasal spine level
Thickness (mm) 4.2 ± 1.1 5.1 ± 1.3 5.0 ± 1.3 p < 0.001
Width (mm) 9.7 ± 1.7 9.7 ± 1.6 9.2 ± 1.6 p > 0.05
Angle relative to the sagittal plane (degree) 101.7 ± 4.7 101.4 ± 4.4 103.8 ± 4.8 p > 0.05
Lower border of the pterygomaxillary junction level
Thickness (mm) 5.9 ± 1.4 6.0 ± 1.3 5.9 ± 1.3 p > 0.05
Width (mm) 8.8 ± 1.7 8.6 ± 1.6 8.2 ± 1.2 p > 0.05
Angle relative to the sagittal plane (degree) 113.4 ± 9.0 115.7 ± 7.0 118.6 ± 8.2 p > 0.05
Other measurement
Distance to the distal aspect of the second molar (mm) 9.2 ± 2.1 9.2 ± 2.1 9.5 ± 1.5 p > 0.05
Angle relative to the maxillary occlusal plane (degree) 98.8 ± 5.1 100.8 ± 5.2 99.0 ± 5.5 p > 0.05

Table 2. Demographic and anatomic variables of the pterygomaxillary junction of cleft side and non- cleft
side from unilateral cleft patients, and also bilateral cleft sides from bilateral cleft patients. All data showed in
mean ± SD. P level = posterior nasal spine level. L level = lower border of the pterygomaxillary junction level.

71%, respectively). For the unfavorable fracture types, among these four groups, the percentage of the maxillary
sinus type was the highest in the Class III group (22%), and the percentage of the pterygoid plate fracture type was
the highest in the cleft group (23%). Different groups were found to have significant association with the separa-
tion types (Chi-square test, p = 0.0013). As for the analysis within the cleft group, the percentage of the pterygoid
plate fracture type was the highest in the cleft side group (36%). Again, different cleft groups were significantly
associated with the separation types (Fisher exact test, p = 0.0387). No severe complications such as hemorrhage
or cranial base fracture were noted in the images or patient’s medical records.

Discussion
Separation of pterygomaxillary junction is a critical step during LeFort I osteotomy to enable complete movement
of the maxilla. However, this step carries risks due to its blind surgical approach. An unfavorable pterygomaxil-
lary separation may cause restricted mobility of the maxillary segment, pterygoid plate fracture, vascular or neu-
ral complications, or blindness. In an attempt to avoid these complications, some authors perform the osteotomy
through the maxillary tuberosity, as described by Trimble35. They asserted that it reduced unfavorable fractures of
the pterygoid plates36 compared to traditional disjunction, and also increases the safety margin to the structures
of the sphenopalatine fossa, reducing the incidence of hemorrhages21, 35. One drawback is that the method can
only be performed if the wisdom teeth are removed. The other one is that the course of the osteotomy, by moving
apart from the pterygopalatine fossa, may come nearer to the descending palatine artery, which is a vulnerable
source of bleeding during the LeFort I osteotomy3. Ligation of the descending palatine artery does not necessarily
jeopardize the blood flow to the maxillary segment, as the gingival blood flow could remain the same with or
without preservation of the artery38. It is advised to carefully perform the osteotomy to prevent injuring the vessel
in order to avoid excessive bleeding and other unfavorable complications39. To achieve this goal, an identification
method for the descending palatine artery during the osteotomy would be necessary.
Full knowledge of the anatomic structures in this region is essential for surgeons performing the procedure.
While the posterior maxilla has been previously studied in normal human skull samples, the specific features
in patients with different malocclusions, if any, were yet to be characterized. This fact seems to be important for
two reasons. First, there is at least one type of abnormal maxillary morphology that has proved to be at high risk
of neurovascular complications16. Second, there seems to be a high inconsistency in results among individuals
from different studies when the same method was used for disjunction, raising doubts whether the anatomy of
the patient itself is a factor. It is worthwhile noting that previous studies regarding the ideal positioning of the
osteotomy only mentioned its angulation in one of the planes of the space, randomly selecting or even neglecting
other planes which may have different anatomic variables. Therefore, results and conclusions can be inadvertently
biased. This is likely the cause of such diverse or even contradictory results in previous studies and makes them
less suitable for comparison.
The main challenge for surgeons during the PMJ separation is finding, without direct vision, the location of
the pterygomaxillary junction from its lateral side. It is typically done by detecting a concavity in the pterygomax-
illary fissure with a curve osteotome. Some patients may have less obvious or undefined PMJ. In this case, it could
be helpful to have an idea of the location by knowing the distance from the most distal aspect of the second molar.
We selected this structure as a landmark and guide because it is the most posterior identifiable hard structure in
the surgical field, and its location is practically constant at an average of 10.9 mm, ranging from 9.3 to 11.6 mm.
Our study showed a significantly shorter distance in the cleft lip/palate groups, averaging 9.3 mm. In the rest of
the groups, we did not find statistically significant differences, averaging 11.0, 11.6, and 11.1 mm for the class I,

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 3

www.nature.com/scientificreports/

Figure 1. The comparisons of different anatomic variables of pterygomaxillary junction. (a) Comparison of the
thickness between the posterior nasal spine level (P level) and the lower border of the pterygomaxillary junction
level (L level), and between the Angle class I, II, III and cleft groups. (b) Comparison of the width between P
level and L level, and between the four groups. (c) Comparison of the angle relative to the sagittal plane between
P level and L level, and between the four groups. (d) Comparison of the distance between the pterygomaxillary
junction and the second molar between the four groups. (e) Comparison of the angle relative to the maxillary
occlusal plane between the four groups. (f) Comparison of the distance between P level and L level between the
four groups. (g) Comparison of the distance between greater palatine foramen and pterygomaxillary junction
between the four groups. (h) Comparison of the thickness between P level and L level, and between cleft side,
non-cleft side and bilateral cleft side. (i) Comparison of the width between P level and L level, and between
cleft side, non-cleft side and bilateral cleft side. (j) Comparison of the angle relative to sagittal plane between P
level and L level, and between cleft side, non-cleft side and bilateral cleft side. (k) Comparison of the distance
between the pterygomaxillary junction and the distal aspect of the second molar between cleft side, non-cleft
side and bilateral cleft side. (l) Comparison of the angle relative to the maxillary occlusal plane between cleft
side, non-cleft side and bilateral cleft side. Statistical analysis was performed by one-way or two-way ANOVA
test. Error bars represent standard deviation. *p < 0.05; **p < 0.01; *** p < 0.001.

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 4

www.nature.com/scientificreports/

Class I Class II Class III Cleft

Total sides 48 106 168 138
Clean-cut type 29 (60%) 73 (69%) 113 (67%) 98 (71%)
Maxillary sinus type 9 (19%) 17 (16%) 37 (22%) 8 (6%)
Pterygoid plate fracture type 10 (21%) 16 (15%) 18 (11%) 32 (23%)

Table 3. Pterygomaxillary junction separation types in postoperative CT at the posterior nasal spine level of
Angle’s classification I, II, III and cleft patients. (Chi-square test, p = 0.0013).

Cleft side Non-cleft side Bilateral cleft

Total sides 52 52 34
Clean cut type 29 (56%) 42 (81%) 27 (79%)
Maxillary sinus type 4 (8%) 3 (6%) 1 (3%)
Pterygoid plate fracture type 19 (36%) 7 (13%) 6 (18%)

Table 4. Pterygomaxillary junction separation types in postoperative CT at the posterior nasal spine level of
cleft side and non-cleft side from unilateral cleft patients, and also bilateral cleft sides from bilateral cleft patient.
(Fisher exact test, p = 0.0387).

class II and class III groups respectively. This represents the distance in a straight line parallel to the sagittal plane
and easy to estimate by the surgeon in the operating room.
The ideal separation line begins laterally in the pterygomaxillary groove, and progresses medially along the
pterygomaxillary junction between the maxilla and the lateral pterygoid process. The lateral to medial course
of the osteotomy in this region is fairly unpredictable and this line creates an angle with the sagittal plane that
is greater than 90° (not perpendicular). One of the previous studies measured a similar angle, concluding it was
around 104° in average in Thai skulls6. Knowing this angulation beforehand guides the surgeon when placing the
osteotome, and therefore avoids osteotomies progressing too anteriorly and approaching the descending palatine
artery, or too posteriorly fracturing the pterygoid plate16. Previous recommendations have been made of using the
osteotome or saw in approximately 90° relative to the sagittal plane28, 30, 40. Our study found that the angle created
by the junction relative to the sagittal plane is, by mean, 100.7°, 104.0°, 102.1° and 102.0° at the posterior nasal
spine level, and 117.1°, 120.0°, 119.3°, 115.1° at the lower border of the PMJ level in patients with the Angle class I,
II, III and cleft lip/palate group respectively. There was no statistically significant difference between non-cleft and
cleft patients. Therefore, it is reasonable to believe that the tip of the instrument used for the separation should be
placed with this angulation, which is greater than the previous recommendation. A limitation of our recommen-
dation is that it may be difficult to evaluate the angulation of the instrument in the operating room, especially with
curved osteotome. However, a rough estimation and practice of the osteotomy direction can be made (Fig. 2).
Our results are in accordance with those from Stajcic41. He proved that by increasing the osteotome angulation
from the conventional 50° to 80°, the incidence of unwarranted fractures of the pterygoid plates decreased signif-
icantly. However, the fractures of both groups were all low-level fractures, and therefore its clinical significance
is questionable.
In the sagittal plane, the distance between the posterior nasal spine and the lower border of the pterygomax-
illary junction were measured. It allowed an idea of how inferior the end of the junction is located, once the
surgeon identifies the posterior nasal spine. It averaged 6.3 mm, 6.3 mm, 5.8 mm and 8.4 mm in class I, II, III and
cleft groups respectively, being the distance in the latter significantly different. Angulation in the sagittal plane
seems to play a role in terms of safety and properly developing the cut. Gulses et al. studied the angulation of the
osteotome in the sagittal plane that caused less unfavorable fractures in 21 Turkish skulls, comparing angulations
of +30°, 0° and −30° relative to the maxillary occlusal plane42. The +30° angulation (osteotome directed inferi-
orly) showed the highest number of high-level fractures. They concluded that positioning the osteotome parallel
to the occlusal plane was the safest choice. Of note, Cheng and Robinson had previously recommended to always
place the osteotome parallel to the maxillary occlusal plane to avoid injuries to the internal maxillary artery28. On
the contrary, Girotto suggested that inferior angulation of the osteotome may minimize the propagation of frac-
tures in a study about atypical fracture patterns of the pterygoid area16. Our study showed that, anatomically, the
pterygomaxillary fissure has an angle greater than 90° to the maxillary occlusal plane (not perpendicular to it) for
patients presenting class I, II or CLP, averaging 108.7°, 93.2° and 99.6° respectively. In class III patients, however,
showed significantly smaller angles, averaging 84.4°. Whether placing the osteotome with the cutting end in this
angulation decreases the occurrence of unfavorable fractures requires further clinical studies. Comparing the P
and L levels, the L level is thicker but not wider. It is assumed that cutting and separating at the L level is more
effective and safer. It is to be noted that the angle is larger in L level than that in P level. This information is helpful
when we are inserting the osteotome and performing the PMJ disjunction.
The reported incidence of serious hemorrhage during LeFort I osteotomy is low. Kramer et al.43 prospec-
tively studied 1000 patients, and described extensive bleeding requiring blood transfusion in 11 (1.1%) patients.
Bleeding can frequently be due to penetrating trauma by an instrument or surgical handpiece, but also to poste-
rior displacement of sharp bone fragments resulting from untoward fractures, injuring the soft tissue in the ptery-
gopalatine fossa. Trauma to the pterygoid plexus is the most likely source of significant bleeding, whereas the

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 5

www.nature.com/scientificreports/

Figure 2. Placement of curved osteotome in the pterygomaxillary junction. Left: 90 degrees to the
sagittal plane. Right: 102 degrees to the sagittal plane, a preferred angulation for smooth separation of the
pterygomaxillary junction.

maxillary artery and its terminal branches are most vulnerable to injure as they course through the pterygopala-
tine fossa. With regards to preventing hemorrhages, awareness of the safety distance from the osteotomy course
to the major vessels is essential. It is not uncommon that during separation, the surgeon directs the osteotome too
anteriorly, approaching the descending palatine artery. This artery can cause profuse bleeding, and most authors
advise to preserve it whenever possible in order to decrease ischemic complications to the mobilized maxilla. We
measured the distance from the pterygomaxillary junction to the greater palatine foramen, where the descending
palatine artery passes through. The average of the distance was 2.8 ± 1.9 (class I), 3.2 ± 0.6 (class II), 3.0 ± 0.3
(class III) and 3.4 ± 1.0 mm (cleft group). Statistically significant difference was only found in the cleft group, who
appears to have the greater palatine foramen more distant to the pterygomaxillary junction than the other groups.
These distances are slightly shorter than those measured by Apinhasmit et al.6. This may be due to differences in
the measurement method; we measured this distance at the axial cut at the level of the posterior nasal spine. On
the other hand, damage to the pterygoid plexus can be minimized by avoiding the occurrence of pterygoid plate
fractures, where bone fragments may be impacted affecting the surrounding soft tissue.
Despite the obvious evidence of increased risk in cleft patients following separation of pterygomaxillary junc-
tion, there has been only one study quantitatively supporting the anatomic differences of the cleft maxilla. Lee
et al. studied the anatomy of the pterygomaxillary junction in cleft lip/palate patients, matching these patients
with non-cleft Angle class III controls44. They concluded that cleft maxilla has greater width and thickness of the
pterygomaxillary junction, and larger greater palatine foramen, with shorter medial pterygoid plate length than
non-cleft class III patients. As mentioned previously, this greater pterygomaxillary thickness has been attrib-
uted to relatively high incidence of complications. This is supported by the study conducted by Wikkeling and
Kopendraaier20, in which the ideal separations occurred in thin, edentulous bones. Our study, in contrast, shows
that cleft patients present thinner junctions but higher incidence of unfavorable fractures at the posterior nasal
spine level in the cleft side, particularly the fractures to the pterygoid plate. None of our patients, however, devel-
oped neurovascular complications related to the posterior maxilla at the time we reviewed the charts. This differ-
ence from the study by Lee et al. has yet to be explained; ethnicity or different surgical techniques for the primary
repair of the cleft during childhood may be factors involved. Of note, the study by Lee et al. showed significant
difference in both sides of the maxilla even in unilateral clefts compared to controls, while in the present study
only the cleft side of the unilateral cleft was different.
Consistent with our results, several other researchers have found a negative correlation between the thick-
ness of the pterygoid junction and the occurrence of the pterygoid plate fracture5, 21, 22. They hypothesized that
when the pterygomaxillary junction is thin, the force from the osteotome may be easily dissipated leading to
unwanted pterygoid plate fracture. One study suggested that pterygoid plate fractures in thin junctions did not
advance upward to the skull base due to its low-level fracture nature22. It is possible that both extreme scenarios,
particularly thick and particularly thin pterygomaxillary junctions, are at high risk. Thick junctions may be more
prone to shatter erratically with an ascending and propagating course through the areas of low resistance, due
to over excessive force exerted by the surgeon. It is still unknown the reason why cleft side in unilateral cleft lip/
palate patients developed thinner junction, while bilateral cleft patients did not show this change. The abnormal
maxillary morphology in the unilateral cleft patients has been suggested by Jiang et al.45, noticing that there was
a significantly shortened maxillary length in the unilateral cleft compared with the normal control group at the
dental level. It may be related to some asymmetric growth pattern that tends to collapse the cleft side, which does
not happen in the bilateral cleft cases.
Syndromic patients with midface hypoplasia could receive a Le Fort III osteotomy and distraction in the
mixed dentition age. The consequences of prior osteotomy through the PMJ as well as the formation of bony
regenerate in this area can present significant challenge at later LeFort I osteotomy. Researchers have studied the
pterygomaxillary region in both cadaveric samples46 and syndromic patients47 in order to better understand the
anatomy in the area. As many of these patients will undergo a Le Fort 1 osteotomy once the growth is complete,
surgeons should expect a thicker and irregular PMJ and a challenging surgical course in these patients.

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 6

www.nature.com/scientificreports/

We classified the pterygomaxillary separation in postoperative CBCT images at the posterior nasal spine level
into three types: clean cut type, maxillary sinus type and the pterygoid plate fracture type. The results showed
higher incidence of pterygoid plate fractures in the cleft patients, with the cleft side being the highest. Of note,
none of our patients presented with neurovascular complications related to the posterior maxilla during the
period comprised in the study. It is known that pterygoid plate fractures present with a higher incidence than
other described complications20, 40. This risk is especifically associated to high-level untoward fractures approach-
ing the skull base, whereas low-level fractures (below the horizontal osteotomy cut) seem to be of little clinical
significance. The latter are undesired mainly because they tend to hamper the advancement of the maxilla due to
the pterygoid muscle attachments to the plates. The injury mechanisms producing neurovascular complications
can be direct, such as associated with bony impingement from an adjacent fracture site, but also by violation of the
pterygopalatine fossa by sharp bone fragments, causing soft tissue damage. Perhaps future studies should focus
on high level fractures. On the other hand, indirect injuries such as traction, compression, or contrecoup may be
sustained during the process of disjunction itself48 which would explain why some cases of complications present
with normal radiographic findings17. Up to this date, the relatively low number of reported cases of complications
does not allow comprehensive study of the many factors involved.
With the introduction of virtual surgery planning, individual characteristics can be identified, and tailored
treatment plans can be employed. However, not every center routinely uses CT imaging prior to the surgery,
and an approximation of the population average anatomy would then be appreciated. An additional aim of this
study was to evaluate the different anatomy in patients with cleft and malocclusion in general terms, which was
accomplished.
A limitation of this study is that the force applied by the surgeon to the junction was not standardized or even
measured. Therefore, it is impossible to evaluate to which extent this was a precipitating factor for occurrence of
atypical fractures. Finally, the results in this study regarding angulation and positioning of osteotome were not
necessarily followed during the surgeries, due to the retrospective nature of this study. Male gender and increased
age has been reported to be a risk factor for the occurrence of pterygoid plate fractures5–7.
In summary, it can be concluded that there are no statistically significant differences in the thickness, width
and angle of the PMJ relative to the sagittal plane between Angle’s classification I, II, III and cleft patients. The
angle relative to the maxillary occlusal plane is smaller in non-cleft class III group compare to the other groups.
Therefore, angulations and distances mentioned above are expected to be the safest when performing the ptery-
gomaxillary disjunction. Differences displayed by cleft patients compared to the other groups are the significantly
shorter distance to the distal aspect of the second molar, significantly longer distance between the posterior nasal
spine and the lower border of the pterygomaxillary junction, and significantly thinner junction in the cleft side of
the unilateral cleft group which may lead to higher incidence of pterygoid plate fracture. Moreover, the posterior
maxilla in the cleft patients is shorter anteroposteriorly. This implies that, in these patients, the vertical cut should
be made closer to the second molar. Also, cleft lip and palate patients show a slight, but significant, greater dis-
tance between the greater palatine foramen and the pterygomaxillary junction, thus preserving a slightly larger
safety margin anteriorly.
Based on the results of this study, it can be concluded that surgeons should note that the cleft group may have
different anatomical structures than normal patient group. Therefore, this study may help surgeons to prepare and
plan the surgery using anatomical measurements as well as characteristics in the specific groups. Furthermore, it
provides useful information in the selection of each group for performing a safer PMJ separation.

Methods
Study designs and patients. A retrospective study was done using the pre- and postoperative CBCT
images of 283 consecutive patients who underwent orthognathic surgery at Chang Gung Craniofacial Center
from January 2012 to December 2015. Patients were classified in groups according to their Angle classification
(class I, class II, class III groups) and presence of cleft lip and palate (cleft group). This cleft group included
patients with unilateral and bilateral cleft lip and palate. Both sides of the maxilla were studied. In the cleft group,
each side was categorized as “cleft side” and “non-cleft side” for the cleft group. A total of 566 sides were evaluated.
For each group, postoperative CBCT at average 1.5 months after the surgery were collected and axial images were
evaluated at the level of the posterior nasal spine. Demographic data were described in Table 1. Exclusion criteria
were previous facial trauma, surgery, tumor, or any other maxillofacial pathologic features other than the repaired
cleft lip and palate in the cleft group. Patients without preoperative records or CBCT images were also excluded
from the study. The study was approved by the Institutional Review Board of Chang Gung Medical Foundation
(No. 201700088B0) and the study was performed following these guidelines. Informed consents were obtained
from the patients or the parents.

Surgical technique. The LeFort I osteotomy was performed by experienced surgeons in the center. For the
pterygomaxillary separation, the pterygomaxillary groove was first identified with the tip of a curve osteotome,
and then the osteotomy proceeded medially by tapping with a mallet. Digital pressure was used to identify pos-
sible remaining sites of resistance; in case these were present, further tapping was performed in the pterygomax-
illary junction. After the disjunction was completed between the maxilla and the pterygoid process, maxillary
downfracture was performed using Röwe disimpaction forceps.

CBCT image analyses. The patient’s CBCT scanning was taken using an i-CAT TM scanner (Imaging
Sciences International, Hatfield, PA, USA) (voxel resolution: 0.4 mm) about 3 weeks prior to the scheduled sur-
gery date. Data were stored in the Digital Image Communications in Medicine (DICOM) format. The DICOM
data were analyzed with a commercial software program, Dolphin 3D (Dolphin Imaging & Management
Solutions, Chatsworth, California, USA) reconstructed into multiple-plane views. Direct measurements were

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 7

www.nature.com/scientificreports/

Figure 3. Landmarks for measurement in the pterygomaxillary junction (PMJ). (a) White square highlighting
the PMJ region at the posterior nasal spine level (P level) in the axial view. The greater palatine foramen (G)
is identified. (b) The thickness at the P level. The perpendicular distance from the most concave point of the
pterygoid fossa to the posterior wall of the maxillary sinus. (c) At the P level, the width was between point
L and point B; the angle relative to the sagittal plane (pointed star) was between the LB line and the sagittal
line; and the distance between the greater palatine foramen and the PMJ was between the center of the greater
palatine foramen and the LB line. (d) The thickness at the lower border of the PMJ level (L level): method same
as (b). (e) At the L level, the width was between point L and point M; and the angle relative to the sagittal plane
(pointed star) was between the LM line and the sagittal line. (f) The angle relative to the maxillary occlusal plane
(pointed star) in the sagittal plane. (g) The distance between the perpendicular extension line created from the
point J and the distal aspect of the root of the second molar. (h) The distance between the P level and the L level.
Definition. Point A: The projection point from the lowest point of the greater palatine foramen to the medial
surface of the PMJ. Point B: The midpoint between point A and C. Point C: The projection point from the most
concave point of the pterygoid fossa to the sagittal line that passes through A. Point L: The most concave point
of the lateral surface of PMJ. Point M: The most medial point of the PMJ in axial view. Point J: The lowest point
of the PMJ in sagittal view.

made with the computer software. In each patient, images from the identical axial level from each CBCT exam-
ination were selected by the examiner. Excel worksheets were designed to list the specific parameters. The data
were then transferred to the computer for statistical analysis.

Anatomical variables. In axial images, two levels were selected to measure the pterygomaxillary region,
one at the posterior nasal spine level (P level) and another one at the lower border of PMJ (L level). The L level
may not be exactly the same on both sides, depends on anatomic variation between sides. Measurements in the
preoperative CBCT were described in Fig. 3.

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 8

www.nature.com/scientificreports/

Figure 4. Pterygomaxillary separation type in postoperative CBCT. (a) Clean cut type: the cutting line within
the PMJ. (b) Maxillary sinus type: part of the posterior wall of maxillary sinus attached to the PMJ after
separation. (c) Pterygoid plate fracture type: pterygoid fracture occurred after PMJ separation.

Pterygomaxillary separation type. Postoperative CBCT axial images were classified into three types
defined as in Fig. 4: (a) Clean cut type: the cutting line right within the PMJ. This cutting type is most desirable for
surgeons. (b) Maxillary sinus type: part of the posterior wall of the maxillary sinus was attached to the PMJ after
separation. The osteotomy was too anterior. (c) Pterygoid plate fracture type: pterygoid fracture occurred after
PMJ separation. The osteotomy was too posterior (Fig. 4).

Error of the measurement. Reproducibility of the measurements was evaluated by comparing differences
between the original and the repeated examinations of 10 randomly selected images in a one-week interval by the
same examiner. The method error was defined as the reproducibility of double determination and it was calcu-
lated as the standard error of measurement. The method error of the repeated linear measurements was 0.3 mm
(differences ranged from 0 to 0.4 mm). The method error in angular measurements was 3° (differences ranged
from 1° to 5°). A statistically significant difference was not detected in neither linear nor angular measurements.

Data analysis. All measurement data were tabulated and separated in the different groups. The mean, stand-
ard deviation (SD) and range for each measurement were calculated. Differences between groups were considered
significant at p < 0.05. Data were analyzed and plotted with GraphPad Prism (version 5.00, La Jolla, USA). With
the two levels of measurement defined as a within-group factor and different groups of malocclusion and cleft lip/
palate group defined as a between-group factor, a two-way ANOVA test was performed to determine the effect
of these factors on the anatomic variables measured in axial CT images. The distance to the second molar, the
distance between the greater palatine foramen and the pterygomaxillary junction, and the distance between P
level and L level were analyzed by one-way ANOVA. When a significant effect was observed, the Tukey test for
one-way ANOVA and Bonferroni test for two-way ANOVA were used as a post hoc test to further characterize
the significance of the specific differences. Statistical significance was set at p < 0.05. The chi-square test and
Fisher exact test were performed to analyze the separation types of each group.

Meeting presentation. The 11th Congress of the Asian Pacific Craniofacial Association, December 1-3,
2016, Nara, Japan.

References
1. Williams, W. G., Lo, L. J. & Chen, Y. R. The Le Fort I-palatal split approach for skull base tumors: efficacy, complications, and
outcome. Plast Reconstruc Surg. 102, 2310–2319 (1998).
2. Belmont, J. R. The Le Fort I osteotomy approach for nasopharyngeal and nasal fossa tumors. Arch Otolaryngol Head Neck Surg 114,
751–754 (1988).
3. Bell, W. H. Le Forte I osteotomy for correction of maxillary deformities. J Oral Surg 33, 412–426 (1975).
4. Melsen, B. & Ousterhout, D. K. Anatomy and development of the pterygopalatomaxillary region, studied in relation to Le Fort
osteotomies. Ann Plast Surg 19, 16–28 (1987).
5. Kanazawa, T. et al. Factors predictive of pterygoid process fractures after pterygomaxillary separation without using an osteotome
in Le Fort I osteotomy. Oral Surg Oral Med Oral Pathol Oral Radiol 115, 310–318, doi:10.1016/j.oooo.2012.04.020 (2013).
6. Apinhasmit, W., Chompoopong, S., Methathrathip, D., Sangvichien, S. & Karuwanarint, S. Clinical anatomy of the posterior maxilla
pertaining to Le Fort I osteotomy in Thais. Clin Anat 18, 323–329, doi:10.1002/ca.20131 (2005).
7. Cheung, L. K., Fung, S. C., Li, T. & Samman, N. Posterior maxillary anatomy: implications for Le Fort I osteotomy. Int J Oral
Maxillofac Surg 27, 346–351 (1998).
8. Carr, R. J. & Gilbert, P. Isolated partial third nerve palsy following Le Fort I maxillary osteotomy in a patient with cleft lip and palate.
Br J Oral Maxillofac Surg 24, 206–211 (1986).
9. Hanu-Cernat, L. M. & Hall, T. Late onset of abducens palsy after Le Fort I maxillary osteotomy. Br J Oral Maxillofac Surg 47,
414–416, doi:10.1016/j.bjoms.2008.10.005 (2009).

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 9

www.nature.com/scientificreports/

10. Lanigan, D. T., Romanchuk, K. & Olson, C. K. Ophthalmic complications associated with orthognathic surgery. J Oral Maxillofac
Surg5 1, 480–494 (1993).
11. Newlands, C., Dixon, A. & Altman, K. Ocular palsy following Le Fort 1 osteotomy: a case report. Int J Oral Maxillofac Surg 33,
101–104, doi:10.1054/ijom.2002.0444 (2004).
12. Watts, P. G. Unilateral abducent nerve palsy: a rare complication following a Le Fort I maxillary osteotomy. Br J Oral Maxillofac Surg
22, 212–215 (1984).
13. Bouletreau, P., Chemli, H., Mortier, J., Freidel, M. & Breton, P. [Severe vascular complications of Le Fort I osteotomy]. Rev Stomatol
Chir Maxillofac 113, 14–18, doi:10.1016/j.stomax.2011.10.003 (2012).
14. Sano, A., Taie, S., Uekita, I., Ueki, M. & Maekawa, N. [Subarachnoid hemorrhage as a complication of Le Fort I osteotomy]. Masui
56, 74–76 (2007).
15. Cheng, H. C., Chi, L. H., Wu, J. Y., Hsieh, T. T. & Pemg, B. Y. Blindness and basal ganglia hypoxia as a complication of Le Fort I
osteotomy attributable to hypoplasia of the internal carotid artery: a case report. Oral Surg Oral Med Oral Pathol Oral Radiol Endod
104, e27–33, doi:10.1016/j.tripleo.2007.01.016 (2007).
16. Girotto, J. A. et al. Blindness as a complication of Le Fort osteotomies: role of atypical fracture patterns and distortion of the optic
canal. Plast Reconstruc Surg. 102, 1409–1421; discussion 1422–1403 (1998).
17. Lo, L. J., Hung, K. F. & Chen, Y. R. Blindness as a complication of Le Fort I osteotomy for maxillary distraction. Plast Reconstr Surg.
109, 688–698; discussion 699–700 (2002).
18. Renick, B. M. & Symington, J. M. Postoperative computed tomography study of pterygomaxillary separation during the Le Fort I
osteotomy. J Oral Maxillofac Surg 49, 1061–1065; discussion 1065–1066 (1991).
19. Robinson, P. P. & Hendy, C. W. Pterygoid plate fractures caused by the Le Fort I osteotomy. Br J Oral Maxillofac Surg 24, 198–202
(1986).
20. Wikkeling, O. M. & Koppendraaier, J. In vitro studies on lines of osteotomy in the pterygoid region. J Maxillofac Surg 1, 209–212
(1973).
21. Dadwal, H., Shanmugasundaram, S. & Krishnakumar Raja, V. B. Preoperative and Postoperative CT Scan Assessment of
Pterygomaxillary Junction in Patients Undergoing Le Fort I Osteotomy: Comparison of Pterygomaxillary Dysjunction Technique
and Trimble Technique-A Pilot Study. J Maxillofac Oral Surg 14, 713–719, doi:10.1007/s12663-014-0720-y (2015).
22. Hwang, K., Lee, D. K., Chung, I. H. & Lee, S. I. Le Fort I osteotomy with sparing fracture of lateral pterygoid plate. J Craniofac Surg
12, 48–52 (2001).
23. Cruz, A. A. & dos Santos, A. C. Blindness after Le Fort I osteotomy: a possible complication associated with pterygomaxillary
separation. J Craniomaxillofac Surg 34, 210–216, doi:10.1016/j.jcms.2006.01.001 (2006).
24. Panula, K., Lovius, B. B. & Pospisil, O. A. The need for orthognathic surgery in patients born with complete cleft palate or complete
unilateral cleft lip and palate. Oral Surg Oral Diagn 4, 23–28 (1993).
25. Rachmiel, A. Treatment of maxillary cleft palate: distraction osteogenesis versus orthognathic surgery–part one: Maxillary
distraction. J Oral Maxillofac Surg 65, 753–757, doi:10.1016/j.joms.2006.08.010 (2007).
26. Ross, R. B. Treatment variables affecting facial growth in complete unilateral cleft lip and palate. Cleft Palate J 24, 5–77 (1987).
27. Turvey, T. A. & Fonseca, R. J. The anatomy of the internal maxillary artery in the pterygopalatine fossa: its relationship to maxillary
surgery. J Oral Surg 38, 92–95 (1980).
28. Cheng, L. H. & Robinson, P. P. Evaluation of a swan’s neck osteotome for pterygomaxillary dysjunction in the Le Fort I osteotomy.
Br J Oral Maxillofac Surg 31, 52–53 (1993).
29. Hernandez-Alfaro, F. & Guijarro-Martinez, R. “Twist technique” for pterygomaxillary dysjunction in minimally invasive Le Fort I
osteotomy. J Oral Maxillofac Surg 71, 389–392, doi:10.1016/j.joms.2012.04.032 (2013).
30. Juniper, R. P. & Stajcic, Z. Pterygoid plate separation using an oscillating saw in Le Fort I osteotomy. Technical note. J
Craniomaxillofac Surg 19, 153–154 (1991).
31. Laster, Z., Ardekian, L., Rachmiel, A. & Peled, M. Use of the ‘shark-fin’ osteotome in separation of the pterygomaxillary junction in
Le Fort I osteotomy: a clinical and computerized tomography study. Int J Oral Maxillofac Surg3 1, 100–103, doi:10.1054/
ijom.2001.0179 (2002).
32. O’Regan, B. & Bharadwaj, G. Prospective study of the incidence of serious posterior maxillary haemorrhage during a tuberosity
osteotomy in low level Le Fort I operations. Br J Oral Maxillofac Surg 45, 538–542, doi:10.1016/j.bjoms.2007.03.006 (2007).
33. Precious, D. S., Morrison, A. & Ricard, D. Pterygomaxillary separation without the use of an osteotome. J Oral Maxillofac Surg 49,
98–99 (1991).
34. Sakai, Y., Kobayashi, S., Sekiguchi, J. & Ohmori, K. New method of endoscopic pterygomaxillary disjunction for a Le Fort Type I
osteotomy. J Craniofac Surg 7, 111–116 (1996).
35. Trimble, L. D., Tideman, H. & Stoelinga, P. J. A modification of the pterygoid plate separation in low-level maxillary osteotomies. J
Oral Maxillofac Surg 41, 544–546 (1983).
36. Lanigan, D. T. & Guest, P. Alternative approaches to pterygomaxillary separation. Int J Oral Maxillofac Surg 22, 131–138 (1993).
37. Hiranuma, Y., Yamamoto, Y. & Iizuka, T. Strain distribution during separation of the pterygomaxillary suture by osteotomes.
Comparison between Obwegeser’s osteotome and swan’s neck osteotome. J Craniomaxillofac Surg 16, 13–17 (1988).
38. Dodson, T. B., Bays, R. A. & Neuenschwander, M. C. Maxillary perfusion during Le Fort I osteotomy after ligation of the descending
palatine artery. J. Oral Maxillofac. Surg. 55, 51–55 (1997).
39. Morris, D. E., Lo, L. J. & Margulis, A. Pitfalls in Orthognathic Surgery: Avoidance and Management of Complications. Clin Plast Surg
34, e17–e29 (2007).
40. Wikkeling, O. M. & Tacoma, J. Osteotomy of the pterygomaxillary junction. Int J Oral Surg 4, 99–103 (1975).
41. Stajcic, Z. Altering the angulation of a curved osteotome–does it have effects on the type of pterygomaxillary disjunction in Le Fort
I osteotomy? An experimental study. Int J Oral Maxillofac Surg 20, 301–303 (1991).
42. Gulses, A. et al. Assessment of the lateral pterygoid plate fractures during Le Fort I osteotomies regarding the angulation of the
osteotome. Stomatologija 16, 61–64 (2014).
43. Kramer, F.-J. et al. Intra- and perioperative complications of the LeFort I osteotomy: a prospective evaluation of 1000 patients. J
Craniofac Surg 15, 971–7– discussion 978–9 (2004).
44. Lee, S. H. et al. Evaluation of pterygomaxillary anatomy using computed tomography: are there any structural variations in cleft
patients? J Oral Maxillofac Surg 69, 2644–2649, doi:10.1016/j.joms.2011.01.004 (2011).
45. Jiang, C., Yin, N., Zheng, Y. & Song, T. Characteristics of Maxillary Morphology in Unilateral Cleft Lip and Palate Patients Compared
to Normal Subjects and Skeletal Class III Patients. J Craniofac Surg 26, e517–523, doi:10.1097/SCS.0000000000002028 (2015).
46. Orra, S. et al. Relevant Surgical Anatomy of Pterygomaxillary Dysjunction in Le Fort III Osteotomy. Plastic and Reconstructive
Surgery 139, 701–709 (2017).
47. Hopper, R. A. et al. Computed Tomographic Analysis of Temporal Maxillary Stability and Pterygomaxillary Generate Formation
following Pediatric Le Fort III Distraction Advancement. Plastic and Reconstructive Surgery 126, 1665–1674 (2010).
48. Polley, J. W. Blindness as complication of Le Fort I osteotomy for maxillary distraction—discussion. Plast Reconstr Surg 109, 699–700
(2002).

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 10

www.nature.com/scientificreports/

Acknowledgements
We would like to thank Dr. Yu-Ray Chen for sharing patients’ image data in the study, Miss Yi-Tan Hung for
collection of image data, and Miss Pei-Ju Lin for statistical analysis. This study was supported by a grant from
the Ministry of Science and Technology (MOST 103–2314-B-182-028-MY3), and a grant from Chang Gung
Memorial Hospital (CRRPG5C0291-3). We would like to thank Dr. Rodney E. Schmelzer for English language
editing.

Author Contributions
Performing image data measurement (Y.P.C., M.B.L.), formulating study method (S.D., K.H.C.), preparation
of data and writing of manuscript (Y.P.C., M.B.L., S.D., K.H.C.), management of image data collection and
measurement (H.H.L.), supervising the study process and editing manuscript (L.J.L.).

Additional Information
Competing Interests: The authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2017

SCieNTifiC REPOrTS | 7: 9953 | DOI:10.1038/s41598-017-10592-8 11