This systematic review was designed and conducted according to the Cochrane Handbook for Systematic Reviews of Interventions¹⁶, and reported according to the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) statement¹⁷. The study protocol was registered in the International Prospective Register of Systematic Reviews (CRD42022352971).

The focused question of this review was: “What is the anatomical position of the lingual nerve with reference to the soft tissue, dental and bony landmarks of the posterior mandible?”

An electronic systematic search of MEDLINE (PubMed) and Embase was conducted by two independent reviewers (P.R.S. and J.R.J.C.) for articles published until 12/9/2022. The results were imported into EndNote reference management software (EndNote ver. 20.4; Clarivate Analytics), merging the search results and removing duplicate records. To identify any additional eligible studies, the reference lists of included studies were screened. The detailed search strategy was recorded in Supplementary Table 1.

This review included studies that report on the position of the lingual nerve with reference to the soft tissue, dental and bony landmarks of the posterior mandible. This encompassed both anatomical studies involving cadaveric dissection, as well as clinical studies employing various imaging modalities or surgical exploration to identify the lingual nerve. Studies were excluded based on the following criteria: (1) in vitro studies, (2) in vivo studies involving animals, (3) case reports or case series with a sample size of fewer than 5 patients, (4) narrative reviews, opinion abstracts, and letters to the editor, and (5) publications in languages other than English.

Study selection was performed in stages by two independent and calibrated reviewers (P.R.S. and J.R.J.C.). The title and abstracts of the retrieved records were screened after which full-text reports were retrieved and reviewed for inclusion based on the above eligibility criteria. Studies without or with unclear abstracts were included for full-text analysis to minimize the exclusion of potentially relevant articles. The agreement between the two reviewers for the title and abstract screening was evaluated using Cohen’s kappa. Any disagreements encountered were resolved through discussion with a third author (S.X.Y.L.).

Two independent reviewers (P.R.S. and J.R.J.C.) extracted data from the main text and tables by using standardized pre-tested electronic data collection forms. All data extracted were confirmed by a third reviewer (S.X.Y.L.). In the event of incomplete/missing data, attempts were made to contact the corresponding authors for clarification. The extracted data included study characteristics (author, year of publication, country), methodological details (type of study, study design, methodology for lingual nerve identification, sample size, and reference points for nerve measurements), and subject characteristics (age, sex, ethnicity, presence of pathology).

The primary outcome of this review was the vertical relationship between the lingual nerve and the dental, hard, and soft tissue landmarks of the posterior mandible. This was quantified as either the prevalence at which the lingual nerve is located above the lingual alveolar crest or the vertical distance between these landmarks and the nerve. The secondary outcome is the horizontal relationship defined either by the prevalence of the lingual nerve contacting the lingual plate or the distance between this landmark and the nerve.

The risk of bias in individual studies was assessed using the Anatomical Quality Assessment (AQUA) Tool¹⁸, by two independent reviewers (S.X.Y.L. and P.R.S.). In cases of disagreements, the risk of bias assessments was resolved by discussions with a third reviewer (J.R.J.C.). The AQUA Tool¹⁸ comprised 5 domains: (1) objectives(s) and subject characteristics, (2) study design, (3) methodology characterization, (4) descriptive anatomy, and (5) reporting of results. Each domain was graded as either high, low, or unclear. Attempts were made to contact the authors of the included studies when clarification was necessary.

The number of lingual nerves presenting above the alveolar crest, and that contacting the lingual plate were recorded as a percentage of the total number of nerves in each study. The reported vertical and horizontal distance between the lingual nerve and the lingual alveolar crest and lingual plate respectively were also recorded as the mean and standard deviation (SD). If the SD was not reported, it was estimated from the standard error.

For both primary and secondary outcomes, individual meta-analyses were performed to estimate the weighted effect sizes with 95% confidence interval (CI). The prevalence was transformed using the arcsine transformation to stabilize the variance. Subgroup analyses were conducted to determine if the weighted effect size were different for the retromolar and third molar regions. The random effects model was used in the meta-analyses to account for heterogeneity among studies. The heterogeneity between studies was assessed using the I² index and Cochran’s Q test. An I² value greater than 50% indicated a high heterogeneity whereas a P<0.1 for the Cochran’s Q test indicated evidence of heterogeneity. All analyses were performed using meta and metafor package in R software (2019; R Foundation for Statistical Computing) (https://www.R-project.org/), at a level of significance of 5% (α=0.05).

The systematic review process is summarized in the PRISMA flowchart.(Fig. 1) The electronic search yielded a total of 2,700 unique records. After screening the titles and abstracts, 47 full-text reports were retrieved for eligibility assessment (inter-examiner agreement: κ=0.942). Thirty-one reports were excluded for the reasons listed in Fig. 1. An additional three studies were identified during the hand search. A total of 18 studies fulfilled the eligibility criteria, and 15 studies were included for meta-analyses.

The characteristics of the 18 included studies^19-36 are summarized in Table 1. These studies were published between 1984 and 2022, reporting on a total of 1,674 nerves. There was a wide geographical distribution with the majority of studies from Asia^{19,22,27,29,31,32,36} and North America^{23,25,30,34,35}. Five out of the 18 studies were clinical studies^{20,21,23,30,34}. The lingual nerve was identified using MRI in three studies^20,21,34, clinical ultrasonography²³, and intra-operative observation during third molar surgery³⁰ in one study each, respectively. Among the five clinical studies, two studies also reported on cadaveric dissection^23,30. In total, 15 studies utilized cadaveric dissection to identify the lingual nerve^{19,22-33,35,36}. In addition, the position of the lingual nerve was identified using radiographic imaging in three studies. The nerve was visualized by painting water-soluble barium on the nerve³¹, or by attaching a wire to the nerve or into its sheath^25,29. The reported position of the lingual nerve and landmarks used in each study are summarized in Tables 2-5.

The lingual nerve was localized using both hard and soft tissue landmarks at the retromolar, third, second, and first molar regions, in five, thirteen, four, and three studies respectively. At the first molar region, the lingual nerve was located 25.20±4.42 mm, 14.38±4.35 mm, and 13.0±4.0 mm apical to the occlusal plane³⁶, alveolar crest¹⁹, and cementoenamel junction²⁵ respectively. At the second molar region, the lingual nerve was more apically positioned, located 17.90±5.26 mm, 12.34±3.16 mm, 11.46±2.98 mm, and 9.6±3.5 mm apical to the occlusal plane³⁶, distolingual attached gingiva²¹, alveolar crest¹⁹, and cementoenamel junction²⁵ respectively. Notably, there were no differences in nerve’s position on the left and right^19,25, and no lingual nerves were found to have a supracrestal location in the first and second molar regions^19,21,25,36.

For the third molar and retromolar regions, the majority of studies evaluated the position of the lingual nerve with reference to the alveolar crest. Meta-analyses were performed to estimate the prevalence of lingual nerves located above this landmark and the distance from the alveolar crest to the nerve. The studies included in each meta-analysis were consistently found to have a high pooled heterogenicity (I2=74% and 99% respectively; Cochran Q test, P<0.01).(Fig. 2) In the vertical plane, 6.3% (95% CI, 1.9%-12.5%) and 8.8% (95% CI, 1.0%-21.7%) of the lingual nerves had coursed above the alveolar crest in the third molar region and retromolar pad regions respectively.(Fig. 2. A) Similarly, the vertical distance between the alveolar crest and the lingual nerve was estimated to be 7.58 mm (95% CI, 4.32-10.84 mm) and 7.70 mm (95% CI, 6.27-9.14 mm) at the third molar and retromolar regions respectively.(Fig. 2. B)

In addition, several alternative hard and soft tissue landmarks were also utilized, including the retromolar pad, the attached gingiva, and the occlusal plane. The lingual nerve was located 9.64±2.98 mm, 10.77±2.76 mm, and 9.30±6.15 mm from the retromolar pad²¹, attached gingiva of the third molar²¹ and its occlusal plane³⁶ respectively.

The horizontal relationship between the lingual nerve and the lingual plate was examined in 10 studies^{20,22,26-30,33-35}. Meta-analyses were performed using these studies to determine the prevalence at which the nerve contacts the lingual plate, and the distance between the two structures. The included studies for the respective meta-analysis had presented with a high pooled heterogenicity (I²=94% and 99% respectively; Cochran Q test, P<0.01).(Fig. 3) In the horizontal plane, 35.2% (95% CI, 13.0%-61.1%) and 19.9% (95% CI, 0.0%-62.7%) of the lingual nerves would contact the lingual plate in the third molar region and retromolar region respectively.(Fig. 3. A) Similarly, the horizontal distance between the two structures was estimated to be 3.43 mm (95% CI, 1.24-5.62 mm), and 3.30 mm (95% CI, –0.27 to 6.87 mm) for the third molar and retromolar regions respectively.(Fig. 3. B)

Fig. 4 summarises the risk of bias assessment of the included studies. In general, the studies were found to be of a low risk of bias for Domains 2, 4, and 5. For Domain 1, 66.7% of studies had a high risk of bias due to incomplete reporting of the subject ages, ethnicity, and sex^{19,24,27-33,35,36}. For Domain 3, 11 studies were found to have a high risk of bias due to inadequate reporting of methodological details required for replicating the study or the lack of examiner calibration or alignment^{22-24,27-30,32,33,35,36}. For Domain 5, one study was identified to have a high risk of bias for incomplete reporting of the sample size and results³⁶.

Most iatrogenic lingual nerve injuries are known to occur in the posterior mandible, particularly in the retromolar and molar regions^7,12,13. It is crucial for clinicians to have a precise understanding of the location of the lingual nerve in these regions to identify high-risk zones that require extra caution during surgeries, mitigating the risks for iatrogenic injury. To provide a clinically relevant perspective to these anatomical findings, the potential course of the lingual nerve was mapped out in this review. In general, the course of the lingual nerve is located apical to the alveolar crest, with a safety margin of approximately 4 to 12 mm as illustrated in Fig. 5. However, the nerve may be located within the lingual soft tissues above the alveolar crest in 8.8% of the retromolar and 6.3% of the partially erupted third molars. Notably, although none of the studies had reported lingual nerves with a supracrestal position at the first and second molar regions, a rare variant was observed where the lingual nerve was present within the retromolar pad of one specimen²². Since clinicians cannot accurately determine the exact location of the lingual nerve peri-operatively, during dentoalveolar surgery, there is a need to adopt conservative precautions to account for these anatomical variants.

Firstly, crestal incisions should be made closer to the center of the alveolar ridge and kept within keratinized tissue away from the retromolar pad to minimize the risk of sectioning the lingual nerve. When considering a distal wedge procedure for pocket reduction or crown lengthening of mandibular second molars in periodontal procedures, the surgical feasibility should be first assessed by evaluating the intraoral distance between the tooth and the retromolar pad, and the radiographic proximity to the ascending ramus. If there is a need to perform an internal bevel incision near the lingual alveolar crest, it is prudent to ensure that the tip of the scalpel is oriented towards and maintains contact with the alveolar crest and not directed towards the lingual tissues. In contrast, at the first and second molar regions, there are substantially lower risks of sectioning the lingual nerve, thus incisions and flap design would be dictated by other surgical considerations.

The position of the lingual nerve within the lingual tissues can also have other implications for implant-related and pre-prosthetic surgeries. The regions analyzed in this review correspond to the first two zones described in the “mylohyoid preservation technique” used for managing an atrophic posterior mandible^13,37. The results of this review substantiate the precautions described in the technique, where the sharp dissection of the lingual tissues is discouraged, and the lingual flap is gently elevated off the retromolar pad with a periosteal elevator. The lingual flap is then separated from the underlying mylohyoid muscle with the use of blunt instruments. This technique maintains the integrity of the periosteum, preventing iatrogenic injury to the lingual nerve and other vital structures.

Furthermore, in the horizontal dimension, the position of the lingual nerve is closely related to the mandible, with 19.9% and 35.2% contacting the lingual plate at the retromolar and third molar regions respectively.(Fig. 3) Thus, considering its proximity to the periosteum, additional caution is necessary when elevating and manipulating the lingual flap since excessive pressure and stretching during flap retraction can also result in iatrogenic injury. It is not recommended to routinely place in a periosteal elevator in the lingual tissues for protection. During the surgical division of impacted third molars, to avoid iatrogenic perforation of the thin lingual cortical plate by the bur, clinicians should first perform an incomplete sectioning with the bur before completing it by rotating a hand instrument in the surgically-created cleft³⁸. This is especially critical for deeply and transversely impacted mandibular third molars which may have pre-existing bony fenestrations of the lingual cortical plate at or near the level of the lingual nerve.

Interestingly, this review has also identified several clinical studies that have employed novel non-invasive modalities to image the position of the lingual nerve^20,21,23,34. These techniques may help clinicians single out patients with lingual nerves with a supracrestal position during pre-surgical evaluation, enabling greater precision during treatment execution and thus mitigating the risk of iatrogenic injury. Additional research will be required to validate their diagnostic accuracies. However, their clinical benefits may be limited to procedures with higher risks for iatrogenic injury such as sagittal split osteotomy in orthognathic surgeries since the above-mentioned precautions are applicable and sufficient for most routine dentoalveolar surgeries.

Although this review has identified the zones of higher risks, its clinical applicability is limited by the variable results reflected by the heterogeneity of the existing literature. Firstly, this review incorporated studies from different geographical locations, involving subjects of different ethnic groups. Their anatomical variations contributed to the observed heterogenicity. Moreover, the included studies involved subjects with a wide age range, thus encompassing varying extents of edentulism, periodontal disease, and other age-related changes. These confounders can affect the position of the alveolar crest, which is the most commonly used landmark used to locate the lingual nerve. Unfortunately, these subject characteristics were not comprehensively reported in the included studies, precluding subgroup analysis and meta-regression, thus highlighting the need for improved reporting in future studies. In addition, separate analyses for dentate and edentulous mandibles should be performed to provide a more accurate depiction of the anatomical position of the lingual nerve.

The extent of impaction and position of the third molar within the mandible is another source of heterogenicity³⁴. Lingually tilted molars would present with a lower alveolar crest, reducing the vertical distance to the lingual nerve. Moreover, considering the course of the lingual nerve, it is also likely that a greater vertical distance will be measured when the third molars are mesially positioned following the early loss of the second molar or when there is sufficient space for its complete eruption. Unfortunately, these confounders were poorly reported in the included studies. This is further confounded by studies that have attempted to estimate the position of the third molar when it is absent^19,26,28,36, and the combined analysis of the third molar and retromolar region as a single entity¹⁴. In the present review, the two zones were clearly distinguished, and subgroup analyses were performed to determine if there were any differences in the position of the lingual nerve. Interestingly, similar mean horizontal and vertical distances were observed. However, the retromolar region presented with a higher frequency for lingual nerves with supracrestal positions, whereas the lingual nerve was more likely to contact the lingual plate at the third molar region. While the differences between the retromolar and third molar regions remained inconclusive due to the heterogenous results, it highlighted the need for clearer reporting of the presence and position of the third molar. Alternatively, future studies can consider reporting the position of the nerve at the retromolar region instead since it would be a more reproducible landmark when the third molar is absent, or when analysing atrophic edentulous mandibles.

Another plausible confounder that explains the heterogeneous vertical distances may be the inclusion of the nerves with supracrestal positions. Notably, it was observed that those studies, demonstrating a higher prevalence of the lingual nerve coursing above the alveolar crest, also reported shorter vertical distances. The shortest mean vertical distance of 2.28 mm was reported by the same study that also observed the highest prevalence (17.6%) of the lingual nerves that coursed above the alveolar crest³⁰. Considering the vertical position of these anatomical variants is different from the majority that courses apical to the alveolar crest. Thus, including these nerves with a supracrestal position when measuring the vertical distance between the nerve and the alveolar crest would skew the measurements towards shorter distances. Future studies should report the respective distances for the nerves that course above and below the alveolar crest, accurately conveying the course and position of the lingual nerve. A similar approach should also be adopted for horizontal measurements and when there are unusual anatomical variations, such as the accessory gingival branch³².