Root-level vulnerability and lingual plate fracture risk in mandibular third molars: A retrospective CBCT-based cross-sectional study

Abstract

Background

Prior studies investigating the relationship between lingual plate (LP) fracture risk and mandibular third molar (M3M) anatomy have produced inconsistent findings. Factors such as age, sex, impaction depth, and buccolingual angulation have shown variable associations with fracture risk. Importantly, M3M rotational angulation has rarely been evaluated. While most research has focused on the middle and apical thirds of the root, one clinical study suggested the cemento-enamel junction may represent a structurally vulnerable site.

Objectives

This study aimed to identify anatomical and demographic risk factors associated with LP fracture across the coronal, middle, and apical thirds of the M3M root.

Methods

Cone beam computed tomography (CBCT) images of 300 mandibular third molars were retrospectively reviewed. Data collected included patient sex, age, tooth side, and spatial alignment. M3M root length was measured and divided into coronal, middle, and apical thirds. The thinnest LP region at each level was recorded, and fracture risk was classified into three categories: low, moderate, and high. Ordinal logistic regression was used to determine significant risk factors.

Results

Fracture risk was significantly lower in the coronal third compared to the middle and apical thirds. Younger patients exhibited higher fracture risk at all levels. Greater root length increased risk at the apical third. Horizontal impaction level II, mesial and horizontal angulations, and buccal inclination were associated with elevated risk, while lingual inclination and rotational angulation reduced risk.

Conclusion

Age, root morphology, impaction depth, and three-dimensional M3M angulations are key predictors of LP fracture risk at distinct root levels.

1. Introduction

Surgical mandibular third molars (M3Ms) removal is the most common surgery performed in oral and maxillofacial specialties. M3Ms are located in the retromolar region and contain several vital structures. Therefore, this treatment is associated with several well-documented complications, including alveolar osteitis, sensory disorders, bleeding, haematoma, and tooth fragment displacement.^1,2 Lingual nerve sensation disorder and tooth fragment displacement are closely linked to lingual plate (LP) integrity.³

The LP is the cortical bone layer separating the M3M root and other important anatomical structures in the oral and maxillofacial regions. The LP is usually thin, and its thickness may change significantly according to some M3M anatomical features.^{4, 5, 6, 7, 8} Therefore, investigating the correlation between LP thickness and M3M anatomy is a priority for preventing fractures and related complications.

Previous studies using multivariable analysis to interpret the correlation between LP fracture risk and M3M anatomy reported heterogeneous results. The impact of age, sex, impaction level, and buccolingual angulation varies and is sometimes contradictory among studies. For instance, Huang et al. (2020) reported that the LP fracture risk did not differ among age groups,⁵ whereas other studies showed that age could be either a protective⁷ or risk factor.^4,8 Furthermore, M3M rotational angulation has not been considered in previous studies. Most studies assessed the risk in the middle and apical thirds of M3Ms; however, a prospective clinical trial found LP thickness at their cemento-enamel junction to be most critical for structural integrity.⁹ Therefore, this study investigated the risk factors associated with LP fractures in impacted M3Ms at the coronal, middle, and apical thirds. The fracture risk was determined based on the LP thickness on cone beam computed tomography (CBCT) images.

2. Materials and methods

2.1. Study setting

This study was conducted using 187 CBCT images stored at the Department of Oral Surgery, Faculty of Dentistry, University of Medicine and Pharmacy in Ho Chi Minh City from October 2024 to March 2025. In accordance with the regulations of the university's Biomedical Research Ethics Committee, the requirements for informed consent and ethics approval were waived due to the retrospective design of the study and the anonymity of the CBCT images. From these CBCT images, 300 M3Ms were used to assess the relationship between anatomical characteristics and LP thickness.

The inclusion criteria were as follows: (1) CBCT images including the following information—sex, date of birth, and date of acquisition; (2) CBCT images that were clear and encompassed the examined region; (3) fully developed M3M apices; and (4) two-rooted M3Ms with clear root bifurcation.

The exclusion criteria were as follows: (1) pathological signs of cysts or tumours related to the M3M region on CBCT images; (2) M3Ms diagnosed with external root resorption; and (3) M3Ms or mandibular second molars (M2M) crowns damaged to an extent that might affect the measurement procedure.

2.2. Data collection procedure

All radiographs were acquired using a CBCT Orthophos SL 3D unit (Dentsply Sirona, Charlotte, NC, United States) by a standard protocol with consistent imaging parameters. The machine was operated at 230–240 V and 50/60 Hz, with a current of 12 mA. Each scan had an acquisition time of 14.1 s and a voxel resolution of 0.3 mm. CBCT images that satisfied the criteria were collected and analysed on a computer with a 1920 × 1080-pixel screen using ImplaStation software (ProDigiDent, Batavia, IL, United States). The study variables were divided into three groups: (1) demographic variables, such as sex and age of the patients; (2) M3M anatomical features, such as side, vertical, and horizontal impaction levels; mesiodistal, buccolingual, and rotational angulation of the M3Ms; and root length; and (3) LP thickness in the coronal, middle, and apical thirds of M3M roots.

On the panoramic view of the CBCT images, the mesiodistal angle was defined as the angle between the long axis of the M3Ms and the adjacent M2Ms. The vertical and horizontal impaction levels of the M3Ms were classified according to the Pell–Gregory classification system. Based on the angle between the vertical axes of the M3M and M2M, M3M mesiodistal angulation was categorised into four groups according to the Winter classification (Fig. 1).

Fig. 1.

Determining vertical and horizontal impaction level according to Pell-Gregory classification and mesiodistal angulation of a mandibular third molar.

To measure the buccolingual angulation, the sagittal plane was adjusted on the coronal window until it passed the M2M long axis. The coronal plane was translated into the M3M's long axis in the sagittal window. The buccolingual angulation is the angle between the M3M and the sagittal planes in the coronal window. If the intersection was on M3M, the angulation coronal side was positive and vice versa (Fig. 2A). M3Ms were classified as lingual, vertical, or buccal inclination if the buccolingual angulation was <-10, −10 to +10, or >+10°.

Fig. 2.

Determining buccolingual (A) and rotational angulation (B) of a mandibular third molar.

To measure the rotational angulation, the axial plane was adjusted to the M2M occlusal surface in the sagittal window. The sagittal plane was rotated parallel to the M2M central groove. The coronal plane was then adjusted to pass through the long axis and centre of the M3Ms in the sagittal window. The axial plane was translated to the M3M occlusal surface. The rotational angulation is the angle between the central groove of the M3Ms and the sagittal plane in the axial window. If the intersection was on the M3M mesial side, the angulation was positive and vice versa (Fig. 2B). M3Ms were classified as non-rotated if the rotational angulation was from −20 to +20°; otherwise, they were rotated.

The M3M's root length and LP thickness were measured in the sagittal and axial window, respectively. On the axial window where the M3M occlusal surface was shown, the sagittal plane was translated perpendicular to the M3M's mesiodistal central groove. The researchers divided the sagittal plane into the centres of the mesial and distal roots. Root length was measured from the cemento-enamel junction to the apex of the sagittal window. The root was then divided into the coronal, middle, and apical thirds. LP thickness was measured at the thinnest position in each third (Fig. 3). If the LP was discontinued or perforated by the M3M's root, the thickness was recorded as 0 mm. After LP thickness was measured, the LP fracture risk was classified as low, moderate, and high with thicknesses ≥1, 0–1, and 0 mm, respectively.

Fig. 3.

Measuring root length of a mandibular third molar and lingual plate thickness at coronal, middle, and apical thirds.

2.3. Statistical analysis

All data were collected and analysed using the JASP software version 0.19.3 (University of Amsterdam, Amsterdam, The Netherlands). Categorical variables are summarised using absolute counts and percentages, whereas numerical variables are described using mean values and standard errors.

A paired-sample t-test and one-way ANOVA were applied to assess differences in LP thickness between the mesial and distal roots and among the three measuring positions of the M3M roots. The Friedman test was used to evaluate differences in LP thickness across the coronal, middle, and apical thirds of both the mesial and distal roots.

The Mann–Whitney U and Kruskal–Wallis tests were used to compare LP thickness between categorical groups. Where significant differences were identified by one-way ANOVA or Kruskal–Wallis tests, and post-hoc Bonferroni corrections were applied for pairwise comparisons.

Pearson's correlation coefficients were calculated to assess the association between LP fracture risk and potential risk factors. Variables with a Pearson's r > 0.25 were included in the predictive model using ordinal logistic regression. Statistical significance was set at p < 0.05.

3. Results

Of the 187 CBCT images, 300 M3Ms were used to measure LP thickness. Table 1 shows the participants’ demographic and radiographic characteristics. LP thickness of the mesial root was significantly greater than that of the distal root at all positions (paired sample t-test, p < 0.001). A one-way ANOVA showed that the LP thicknesses of the three parts of both roots were significantly different (p = 0.001). The post-hoc Bonferroni test revealed that the LP thickness of the apical third was significantly lower than that of the coronal third for both roots (p < 0.05). Fig. 4 shows the percentage of LP fracture risk classifications for the M3Ms. The Friedman test showed that LP risk classification differed significantly among the three positions (p < 0.001). The post-hoc Bonferroni test indicated that the risk in the coronal third was significantly lower than those in the middle and apical thirds (p = 0.006 and p < 0.001, respectively).

Table 1.

Patient demography and anatomical features of mandibular third molars of the study sample (n = 300).

Characteristics
Age, mean (SD), year	33.26 (11.80)
Sex, n (%)
Female	187 (62.33)
Male	113 (37.67)
Side, n (%)
Left	152 (50.67)
Right	148 (49.33)
Vertical impaction level, n (%)
A	230 (76.67)
B	59 (19.67)
C	11 (3.67)
Horizontal impaction level, n (%)
I	154 (51.33)
II	135 (45.00)
III	11 (3.67)
Mesiodistal angulation, n (%)
Vertical	105 (35.00)
Mesial	115 (38.33)
Horizontal	52 (17.33)
Distal	28 (9.33)
Buccolingual angulation
Vertical	205 (68.33)
Lingual	62 (20.67)
Buccal	33 (11.00)
Rotation
Non-rotated	251 (83.67)
Rotated	49 (16.33)
Lingual plate thickness of mesial root
Coronal third	1.06 (1.06)
Middle third	0.99 (1.00)
Apical third	0.89 (0.95)
Lingual plate thickness of distal root
Coronal third	0.80 (0.78)
Middle third	0.70 (0.80)
Apical third	0.73 (0.91)

Fig. 4.

Distribution of lingual plate fracture risk at coronal, middle, and apical thirds.

Table 2 shows the M3M LP thickness according to the vertical and horizontal impaction levels and mesiodistal, buccolingual, and rotational angulations. For the vertical impaction level, no significant difference in LP thickness was found between the M3Ms at different levels. The M3M LP thickness decreased gradually from the coronal to apical thirds; however, significant differences were found only in the mesial root of M3Ms at levels A and B. For the horizontal impaction level, the LP of M3Ms was significantly different between the three levels—highest in level I and lowest in level II. Fig. 5 shows the results of the post-hoc Bonferroni test used to compare pairs of root thirds according to the vertical and horizontal levels. Significant differences in LP thickness were found among the root thirds, two roots, and impacted levels of M3Ms. For mesiodistal angulation, mesial inclination M3Ms have significantly thinner LP than their counterparts. The LP gradually thinned from the coronal to apical thirds; however, it was only significant in the mesial root of the mesial inclination M3Ms (p = 0.002). For buccolingual angulation, LP thickness significantly decreased from the lingual to the vertical to the buccal inclination of M3Ms. This reduction was significant in the middle and apical thirds. From the coronal to apical thirds, the LP thickness in the vertical and buccal inclination M3Ms was also significantly decreased, whereas it was increased in the lingual inclination. For rotational angulation, although almost no significant difference in LP thickness was observed when assessing position, the LP thickness of rotated M3Ms significantly decreased from the coronal to apical thirds. Fig. 6 presents the results of Mann-Whitney and post-hoc Bonferroni tests to reveal significant differences between mesiodistal, buccolingual, and rotational angulations.

Table 2.

Comparison of lingual plate thickness of mandibular third molars at each third between vertical and horizontal impacted levels, mesiodistal, buccolingual, and rotational angulation (n = 300).

		Mesial root				Distal root
		Coronal third	Middle third	Apical third	P∗	Coronal third	Middle third	Apical third	P∗
Vertical impaction level	A	1.06 ± 0.91	1.01 ± 0.96	0.95 ± 0.94	0.007	0.77 ± 0.77	0.70 ± 0.82	0.73 ± 0.90	0.062
	B	1.06 ± 1.33	0.93 ± 1.14	0.68 ± 0.85	0.015	0.85 ± 0.71	0.66 ± 0.73	0.63 ± 0.84	0.185
	C	0.99 ± 2.06	0.82 ± 1.00	0.86 ± 1.41	0.576	1.15 ± 1.20	0.99 ± 0.78	1.36 ± 1.44	0.974
	P∗∗	0.073	0.291	0.054		0.354	0.307	0.209
Horizontal impaction level	I	1.30 ± 0.93	1.25 ± 0.98	1.15 ± 1.20	0.022	0.95 ± 0.71	0.85 ± 0.74	0.89 ± 0.85	0.122
	II	0.79 ± 1.14	0.70 ± 0.93	0.99 ± 0.78	0.035	0.60 ± 0.77	0.51 ± 0.84	0.52 ± 0.93	0.071
	III	0.92 ± 0.85	0.83 ± 0.97	1.36 ± 1.44	0.500	1.18 ± 1.21	0.98 ± 0.84	1.17 ± 1.06	0.643
	P∗∗	<0.001	<0.001	<0.001		<0.001	<0.001	<0.001
Mesiodistal angulation	Vertical	1.36 ± 0.98	1.31 ± 0.97	1.27 ± 0.96	0.555	0.92 ± 0.72	0.85 ± 0.75	0.92 ± 0.87	0.729
	Mesial	0.98 ± 1.16	0.89 ± 1.04	0.71 ± 0.86	0.002	0.72 ± 0.68	0.58 ± 0.67	0.61 ± 0.80	0.016
	Horizontal	0.55 ± 0.72	0.49 ± 0.66	0.50 ± 0.78	0.546	0.67 ± 0.83	0.61 ± 0.64	0.65 ± 0.91	0.904
	Distal	1.20 ± 1.04	1.11 ± 0.98	0.97 ± 1.03	0.129	0.90 ± 1.20	0.80 ± 1.46	0.68 ± 1.37	0.079
	P∗∗	<0.001	<0.001	<0.001		0.021	0.009	0.003
Buccolingual angulation	Vertical	1.04 ± 0.99	0.97 ± 1.00	0.86 ± 0.88	0.002	0.74 ± 0.70	0.61 ± 0.66	0.60 ± 0.71	<0.001
	Lingual	0.96 ± 0.82	1.11 ± 0.94	1.24 ± 1.18	0.127	0.95 ± 1.03	1.09 ± 1.14	1.33 ± 1.32	0.006
	Buccal	1.36 ± 1.68	0.87 ± 1.06	0.46 ± 0.57	<0.001	0.86 ± 0.68	0.54 ± 0.59	0.40 ± 0.58	0.001
	P∗∗	0.901	0.221	0.002		0.188	<0.001	<0.001
Rotational angulation	Non-rotated	0.98 ± 0.95	0.89 ± 0.85	0.84 ± 0.85	0.009	0.77 ± 0.71	0.67 ± 0.71	0.72 ± 0.82	0.099
	Rotated	1.46 ± 1.44	1.46 ± 1.47	1.17 ± 1.30	0.018	0.93 ± 1.06	0.86 ± 1.17	0.80 ± 1.31	0.012
	P∗∗∗	0.030	0.016	0.259		0.502	0.382	0.530

Note. P∗ = Friedman test; P∗∗ = Kruskal–Wallis test; P∗∗∗ = Mann–Whitney test.

Fig. 5.

Lingual plate thickness of mandibular third molar according to vertical and horizontal impaction levels.

Fig. 6.

Lingual plate thickness of mandibular third molar according to mesiodistal, buccolingual, and rotational angulation.

Table 3 presents the correlations between LP fracture risk and demographic and radiographic factors. Except for sex, all the other factors showed potential relationships (p < 0.25) with at least one of the three measured positions of the M3M's root. These factors were used for ordinal regression analysis of LP fracture risk factors. In Table 4, the model fitting formation, goodness-of-fit, pseudo-R-square, and parallel lines of the prediction of LP fracture risk for the three root thirds are presented. Generally, the ordinal regression analysis results showed that all models were appropriate and valid for predicting LP fracture risk. Table 5 reports the effects of independent factors on LP fracture risk. A decrease in patients' age (in years) was related to an increase of LP fracture risk at all three root thirds: coronal third (OR = 1.026; 95% CI, 1.002–1.051), middle third (OR = 1.030; 95% CI, 1.006–1.055), and apical third (OR = 1.030; 95% CI, 1.005–1.055). An increase in root length was associated with a decrease in LP fracture risk in the apical third (OR = 1.279; 95% CI, 1.056–1.550). Regarding horizontal impaction level, M3Ms at level II had higher odds of LP fracture risk compared with those at level I—3.040 times in the coronal third (95% CI, 1.520–6.098) and 2.353 times in the middle third (95% CI, 1.185–4.673). Mesiodistal angulation was also a significant factor related to LP fracture risk in the apical third. The odds of mesial and horizontal M3Ms being classified as having a higher LP fracture risk were 1.961 (95% CI, 1.080–3.571) and 3.717 (95% CI, 1.570–8.850) times that of vertical fractures, respectively. Lingual inclination M3Ms were less likely to have an LP fracture risk with OR = 0.416 (95% CI, 0.230–0.751) in the middle third and OR = 0.224 (95% CI, 0.121–0.414) in the apical inclination when compared with vertical M3Ms. Finally, the odds of rotated M3Ms having a lower LP fracture risk in the middle third were 0.370 (95% CI, 0.230–0.751) times those of their counterparts.

Table 3.

Correlation between lingual plate fracture risk at root thirds with demographic and anatomy factors (n = 300).

Factors	Coronal third		Middle third		Apical third
	Pearson's r	P	Pearson's r	P	Pearson's r	P
Age	−0.273	<0.001	−0.308	<0.001	−0.319	<0.001
Sex	−0.034	0.558	0.038	0.514	0.045	0.442
Root length	−0.034	0.558	−0.090	0.122	−0.173	0.003
Vertical impaction level	0.091	0.114	0.119	0.039	0.097	0.092
Horizontal impaction level	0.291	<0.001	0.272	<0.001	0.252	<0.001
Mesiodistal angulation	0.189	<0.001	0.204	<0.001	0.233	<0.001
Buccolingual angulation	−0.106	0.067	−0.011	0.846	−0.026	0.649
Rotational angulation	−0.081	0.162	−0.116	0.045	−0.004	0.944

Table 4.

Model fitting test results of ordinal logistic regression analysis for predicting models of lingual plate fracture risk for three root thirds (n = 300).

	Coronal third	Middle third	Apical third
Model fitting formation
χ2(12)	60.368	68.638	79.899
P	<0.001	<0.001	<0.001
Goodness-of-fit
Pearson χ2(582)	640.254	597.517	596.436
P	0.047	0.319	0.330
Deviance χ2(582)	560.585	558.905	561.121
P	0.731	0.747	0.726
Pseudo R-square
Cox and Snell	0.182	0.205	0.234
Nagelkerke	0.280	0.233	0.264
McFadden	0.097	0.109	0.124
Test of parallel lines
χ2(12)	16.029	6.860	15.950
P	0.190	0.867	0.194

Table 5.

Ordinal logistic regression analysis of risk factors of lingual plate fracture risk (n = 300).

Model	Variable	B	SE	Wald	P
Coronal third	Moderate	1.359	1.127	1.455	0.228
	High	−1.238	1.128	1.205	0.272
	Age	0.026	0.012	4.518	0.034
	Root length	−0.019	0.096	0.039	0.844
	Vertical impaction levela
	B	0.371	0.344	1.165	0.280
	C	−0.261	0.675	0.149	0.699
	Horizontal impaction levelb
	II	−1.112	0.354	9.864	0.002
	III	−0.052	0.724	0.005	0.942
	Mesiodistal angulationc
	Mesial	−0.430	0.308	1.944	0.163
	Horizontal	−0.723	0.430	2.831	0.092
	Distal	−0.255	0.420	0.368	0.544
	Buccolingual angulationd
	Lingual	0.257	0.297	0.751	0.386
	Buccal	0.707	0.388	3.316	0.069
	Rotational angulatione
	Rotated	0.561	0.318	3.101	0.078
Middle third	Moderate	3.084	1.144	7.268	0.007
	High	0.573	1.128	0.258	0.612
	Age	0.030	0.012	5.831	0.016
	Root length	0.085	0.096	0.778	0.378
	Vertical impaction levela
	B	0.047	0.346	0.019	0.892
	C	−0.492	0.698	0.497	0.481
	Horizontal impaction levelb
	II	−0.855	0.350	5.972	0.015
	III	−0.010	0.725	0.000	0.989
	Mesiodistal angulationc
	Mesial	−0.292	0.307	0.908	0.341
	Horizontal	−0.789	0.434	3.305	0.069
	Distal	−0.372	0.418	0.789	0.374
	Buccolingual angulationd
	Lingual	0.878	0.302	8.457	0.004
	Buccal	−0.517	0.390	1.758	0.185
	Rotational angulatione
	Rotated	0.995	0.322	9.539	0.002
Apical third	Moderate	4.567	1.168	15.288	<0.001
	High	2.590	1.146	5.109	0.024
	Age	0.029	0.012	15.288	<0.001
	Root length	0.246	0.098	5.805	0.016
	Vertical impaction levela
	B	0.070	0.351	0.040	0.841
	C	0.147	0.679	0.040	0.829
	Horizontal impaction levelb
	II	−0.414	0.347	1.426	0.232
	III	−0.256	0.714	0.128	0.720
	Mesiodistal angulationc
	Mesial	−0.674	0.305	4.899	0.027
	Horizontal	−1.314	0.440	8.921	0.003
	Distal	−0.746	0.417	3.197	0.074
	Buccolingual angulationd
	Lingual	1.497	0.314	22.775	<0.001
	Buccal	−0.658	0.398	2.730	0.098
	Rotational angulatione
	Rotated	0.327	0.319	1.054	0.305

Note. B, unstandardized coefficient; SE, standard error; ^a reference = A; ^b reference = I; ^c reference = vertical; ^d reference = vertical; ^e reference = non-rotated.

4. Discussion

The LP plays a vital role as a hard tissue barrier between the M3Ms and other important anatomical structures, such as the lingual nerve, sublingual space, and submandibular space. The roots of M3Ms are closely related to LP integrity and can compromise it in some situations. Fracture of the LP and displacement of M3Ms fragments may lead to serious complications that can permanently affect patient health. Previous studies have investigated the LP thickness in different M3M anatomical features to identify risk factors for LP fractures.^4,10-15 However, most of them used bivariate analysis or included only mesioangular/horizontal M3Ms, which did not adjust for the impact of several different features of an impacted M3M.^10,13-17 The results indicated that patient age, horizontal impaction level, and mesiodistal, buccolingual, and rotational angulations were significant factors that could affect LP fracture risk. Interestingly, patient age, lingual inclination, and rotated M3Ms are protective factors that reduce LP fracture risk.

The results showed that the LP thickness of the mesial root was significantly higher than that of the distal root at all assessed positions. The mean values of LP thickness varied from 0.70 to 1.03 mm, and LP fracture risk gradually increased from the coronal to apical parts of the M3M's root. Similarly, most of the previous studies indicated that LP thickness was notably thin, with the reported values ranging from 0.65 to 1.42 mm.^{4,7,10, 11, 12, 13, 14} Specifically, Cardoso et al. (2025) revealed that the thinnest LP position was frequently in the distal root and middle third of the M3M.⁹ Although a few studies have reported greater LP thickness, their measurement protocols differed in direction and landmark determination.^9,15

In this study, age was found to have a positive correlation with LP thickness in the M3M regions, and an increase in age helped reduce LP fracture risk in all root thirds. This finding is consistent with previous observations reported in the literature. Halder et al. (2023) observed that older patients tended to have a thicker LP at the M3M apex than younger ones.⁷ Wang et al. (2016) also showed that 25–35-year-old patients had a lower rate of LP perforation than 18–25-year-old patients.⁴ Although this relationship was not statistically significant in the study by Huang et al. (2020), the same tendency was confirmed by univariate analysis.⁵ These findings reflect age-related changes in the LP thickness, which may result from bone remodelling and mineralisation. A postmortem CT study reported slight increases in lingual alveolar bone thickness with age, particularly in the retromolar region, whereas in other areas, age effects were minimal or absent.¹⁶ That is, LP thickness increases with age, suggesting a lower risk of complications—such as LP fracture or lingual nerve injury—during third molar extraction in older patients.

Although the vertical impaction level did not show a significant effect, the horizontal impaction level, as defined by the Pell–Gregory classification, was significantly associated with LP fracture risk. For horizontal impaction level, the previous CBCT–based studies also confirmed that M3Ms classified as Pell–Gregory class I and II were prone to have LP perforation.^8,17,18 Class III M3Ms are deeply impacted in the ramus; therefore, the LP is frequently thick and has a low risk of fracture. In contrast, Huang et al. (2020) did not find any correlation between LP fracture risk and the horizontal impaction level; however, their landmarks for measuring LP thickness were different from ours.⁵ For the vertical impaction level, this study's findings were consistent with those of the previous studies that showed no greater M3M risk in class B or C LP fractures.^4,5,7 Thus, M3Ms classified as Pell–Gregory class II are at greater LP fracture risk than those in class I.

Mesial and horizontal M3Ms are more likely to be at high LP fracture risk, especially at the apical third. This finding is consistent with previous studies that implied a significant reduction in LP thickness and a high LP perforation percentage in the mesial and horizontal M3Ms, especially at the mid-root and apex levels.^{5,7,8,12,18, 19, 20} Mesial and horizontal impacted M3Ms roots were in a more posterior region than the vertical and distal roots. Ueda et al. (2023) indicated an abrupt decrease in mandibular width from the midpoint between the mental foramen and the antegonial notch to the mandibular angular apex, where M3Ms are usually located.²¹ In other words, a more posterior M3M apex increases the likelihood of a thin or root-perforated LP.

Lingual inclination was a protective factor against LP perforation, whereas buccal inclination did not significantly reduce LP thickness. Some CBCT investigations have also demonstrated that vertical M3Ms have a greater chance of perforating the LP than lingually inclined ones.^10,12 In contrast, larger-scale studies did not find buccolingual angulation to be a consistent predictor when controlling for confounding variables.^5,7 Mahabob et al. (2024) revealed that M3M roots were much closer to the LP (1.1–2.1 mm) than the buccal plate (3.7–4.3 mm).²² In other words, the lingual aspect of an M3M offers less space and denser bone than the buccal side. Therefore, we hypothesised that lingually inclined M3Ms exhibit root inclination toward the buccal cortical plate, whereas buccally inclined molars demonstrate crown inclination toward the LP. Accordingly, lingually inclined M3Ms were associated with a significantly decreased risk of LP fractures, whereas the increase in risk observed in buccally inclined cases was not statistically significant.

To date, no studies have identified M3M root length and rotational angulation as significant factors affecting lingual cortical plate thickness. Interestingly, the results showed that M3M rotation helped reduce LP fracture risk, especially in the middle third. Additionally, M3Ms with shorter root are more likely to have higher LP fracture risk. M3Ms are typically positioned within the retromandibular triangle, where the mandibular body diverges laterally at a horizontal angle of approximately 20°–30° before transitioning into the ramus.²³ Therefore, the rotated M3Ms appeared to remain intraosseous and exhibited a lower tendency to compromise the LP integrity. Regarding correlation between the root length and LP fracture risk, we suggest two hypotheses. First, shorter-rooted M3Ms might appear to have greater buccolingual root dimensions, possibly as an adaptive morphological feature to maintain stability despite reduced root length. Second, shorter-root is a consequence of contact between M3Ms root and lingual cortical plate during root formation. However, these relationships have not been systematically evaluated in the literature and warrants further investigation.

This CBCT-based study has several advantages in terms of strengthening the results. CBCT images were obtained following a single parameter by a radiographic technician to ensure unique quality. M3Ms were evaluated in multiple dimensions to identify potential factors related to M3Ms that may affect LP thickness. However, this study has some limitations. First, the study population was derived from a single centre and may not reflect broader anatomical variations across age, ethnicity, or skeletal class. The distribution of Pell–Gregory classification levels for M3Ms varies across different ethnic populations, as reported in Indian and Japanese cohorts and observed in the present Vietnamese sample.^7,8 Second, there is no universally accepted standard for quantifying M3M rotational angulation, and measurement variability may affect the interpretation of its clinical impact.²⁴ To our knowledge, this is the first study to evaluate M3M rotation impact, defined as a deviation greater than 20° relative to the adjacent M2M, on LP thickness. Accordingly, the classification threshold used for the rotational angulation in this study warrants further validation in future research. Finally, this study focused only on cortical thickness and did not evaluate bone density, bone volume, or three–dimensional morphology, which may also influence and more comprehensively reflect the LP fracture risk. Some studies reported that mandibular morphology and density might also affect LP thickness.^4,7 Therefore, clinicians should interpret and apply these findings with consideration of age-related physiological changes in the lingual plate, including cortical thickness, morphology, and bone density.

5. Conclusions

The LP fracture risk in the M3Ms gradually increased from the coronal to the apical third. Multiple patient demographic and M3M anatomical factors influence this risk. Younger age, short root, horizontal impaction level II, mesial and horizontal angulation significantly increased the fracture risk at different root levels.

Patient's/Guardian's consent

Not applicable.

Author contributions

Son Hoang Le: Conceptualization, Methodology, Formal analysis and investigation, Writing - original draft preparation, Writing - review and editing.

Bich-Ly Thi Nguyen: Conceptualization, Methodology, Writing - original draft preparation, Writing - review and editing.

Ethical clearance

Not required.

Human and animal statement

The work has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans (https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/); EU Directive 2010/63/EU for animal experiments (http://ec.europa.eu/environment/chemicals/lab_animals/legislation_en.htm).

Sources of funding

This research did notreceive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.