In this study, the association between a surgeon’s choice of mandibular third molar coronectomy and the Pell and Gregory or Winter classification system was assessed. Panoramic radiographs were examined from patients whose preoperative radiographs revealed a close relationship between the mandibular third molar and the IAN. Use of the Pell and Gregory classification system was significantly related to the surgeon’s decision to perform a coronectomy (p = 0.002). Class III (78.6%) and position B (67.4%) were the most common classifications in cases of coronectomy. The Winter classification system was not associated with the surgeon’s choice of procedure (p = 0.425). Following the initial analyses, a post hoc test was performed to probe the origin of statistically significant associations between Pell and Gregory classifications and surgeon assessment. This analysis identified two pairs of factors that were statistically significant: classes II (58.9%) and III (78.6%) and positions A (54.0%) and B (67.4%; both p < 0.0167). Other classification pairs were not statistically significant after Bonferroni correction, indicating that the initial significant association was not solely attributable to class III and position B impactions.
To the best of our knowledge, no studies have investigated choice of surgical procedure in association with the Pell and Gregory or Winter classification as evaluated on a preoperative panoramic radiograph. Most previous work using these classification systems in relation to a coronectomy has focused on root migration of mandibular third molars following a coronectomy. A few studies have used these classification systems to assess the diagnostic value of panoramic radiographs in predicting a true relationship between the mandibular third molar and the IAN [16, 17, 19, 20]. In those investigations, the impaction pattern was compared with a gold standard, either a CT or cone-beam CT scan. Other studies have compared the impaction pattern with a more clinical outcome, such as IAN damage or IAN exposure [7, 11, 15, 21], although the surgical procedure was limited only to surgical removal of the mandibular third molar. In the present study, the impaction pattern was compared with the surgeon’s choice of procedure. To limit the possibility of IAN injury, after reviewing CT scans, the surgeon could decide on a coronectomy rather than surgical removal. Although the surgeon’s ultimate assessment was based on CT scan data, the current findings cannot be directly compared with those of earlier studies. Given the absence of directly comparable studies, however, a comparison is given below to contextualize the present findings.
Monaco et al. assessed an association between the Pell and Gregory classification and the topographic relationship between the third molar and the IAN [16, 20], evaluating the accuracy of the radiographic markers by comparing the panoramic radiograph with a CT scan [16]. They found a significant association of the space between the second molar and the ramus (class I/II/III), as seen on the panoramic radiograph, and a true relationship between the third molar and the mandibular canal, as determined by the CT scan (p = 0.03). In comparison, in the present study using panoramic radiographs, impaction patterns were compared by the surgeon’s choice of procedure, ultimately based on CT scan data. In cases of a true relationship, class III was most common, a consistency among the findings of these studies. Monaco et al. [16] also noted that as the amount of coverage by the ramus increased, the frequency of a true relationship increased. The present study found a significant association between the surgeon’s decision and the space between the second molar and the ramus (p = 0.005). Nunes et al. [20], in contrast, found no significant association of the space between the second molar and the ramus and intimate contact with the mandibular canal (p = 0.57). In that study, panoramic radiographs were compared with cone-beam CT scans to evaluate predictive panoramic signs.
Nakagawa et al. [19] and Nunes et al. [20] found no significant association between impaction depth (position A/B/C) and intimate contact with the mandibular canal. However, Monaco et al. [16] reported an association between impaction depth and a true relationship with the IAN (p = 0.02), which is again consistent with the current study. The similarities between the study groups in Monaco et al. [16] and the current work could explain the consistency, and both investigations employed radiographic signs as selection criteria, whereas Nunes et al. [20] did not. Worth noting is that these previous studies had sample sizes ranging from 73 to 148 molars [16, 19, 20], whereas the present study used a total of 813 mandibular third molars. As such, any hypothetical inaccuracy in classifying third molars would have affected study outcomes more with a lower sample size.
In the present study, a post hoc analysis revealed that class III (p = 0.001) and position B (p < 0.001) were primarily responsible for the overall association between the surgeon’s choice and the two categories of the Pell and Gregory classification. Some studies have used bone coverage to define “impaction depth” instead of using the space between the second molar and the ramus according to the Pell and Gregory classification [7, 15, 21]. The term “impaction depth” thus is defined differently in these studies than in the current work, having been used to classify the amount of bone coverage (erupted, partly erupted, unerupted). The results of these studies have suggested that the amount of bone covering the molar correlates with postoperative sensory changes.
In comparison to the other classes and positions within the Pell and Gregory classification, class III and position C can be considered as reflecting higher amounts of bone coverage [16, 19, 20]. According to Carmichael and McGowan [21] and Kipp et al. [15], neurosensory changes are seen most often when the third molar is entirely covered with bone. Valmaseda et al. [7] found that intraosseous impaction was linked to a higher risk of IAN injury, although not significantly so (OR = 2.82, p = 0.088). These results suggest that increased bone coverage could lead to more coronectomy indications, although such an association did not emerge in the present study (Tables 3–5).
Use of the Winter classification of a mandibular third molar has led to conflicting results regarding associations with a true relationship with the IAN. Carmichael and McGown [21], Kipp et al. [15], and Nunes et al. have reported such an association. Nakagawa et al. [19] and Valmaseda et al. [7], as well as the current study (p = 0.425), however, found none. The observed frequencies for the different inclinations in these studies also are inconsistent, but a bias towards a certain distribution could not be identified. The inclination of an impacted mandibular third molar may have no influence on the surgeon’s decision to perform a coronectomy, as was observed in the present study (Table 6).
This study had some limitations. Only one reviewer evaluated the preoperative panoramic radiographs, although consistency was determined through an analysis using kappa statistics to reflect intra-rater reliability. The kappa values ranged from good to very good, indicating a consistent measurement at the two time points. The validity of this single reviewer, i.e., inter-rater reliability, could not be evaluated in the absence of other reviewers, leaving the possibility that the reviewer was consistently wrong, without a way to verify the accuracy.
For study inclusion, involved molars had to be in a close relationship with the IAN, as indicated when at least one of five radiographic signs associated with increased risk of IAN injury were visible on the preoperative panoramic images. Darkening and narrowing of the root, interruption of the white line, and narrowing and diversion of the canal were all regarded as signs of study eligibility [10, 11, 18]. Most studies used for comparison in the current work focused on a sample of patients who required removal of impacted mandibular third molars, without considering preoperative radiographic signs [7, 15, 20, 21]. No selection criteria related to risk of IAN injury were used in these studies. Not using radiographic signs for selection criteria can affect the distribution of the sample, yielding relatively fewer high-risk patients, but in the present study, only high-risk patients were eligible. This difference may have influenced outcomes in earlier studies, making comparison with the current study more difficult.
A question arises regarding the clinical utility of knowing the extent to which the Pell and Gregory or Winter classification can affect a surgeon’s decision to perform a mandibular coronectomy. Relying only on impaction pattern and foregoing additional 3D imaging would not be advised [22]. However, limiting additional imaging would result in lower radiation exposure and decrease financial costs compared with panoramic imaging [23]. A poor diagnosis, on the other hand, could result in serious complications, such as postoperative neurosensory impairments [8]. Knowing the association of mandibular third molar impaction patterns and the risk for IAN injury could help a clinician determine whether to use additional 3D imaging, supporting decisions about surgical techniques to limit the risk of IAN injury.
In conclusion, this study showed a relationship between the Pell and Gregory classification and a surgeon’s decision to perform a mandibular third molar coronectomy. Coronectomy was preferred over surgical removal more frequently for molars in class III and position B. The Winter classification of mandibular third molars was not associated with the surgeon's choice. Future studies should assess the accuracy of the Pell and Gregory classification in predicting a true relationship with the mandibular canal.