Clinical and radiographic outcomes of second molars with external root resorption following impacted third molar removal: a retrospective CBCT-based cohort study

Abstract

Objective

To evaluate the clinical and radiographic outcomes of second molars (M2s) affected by external root resorption (ERR) following impacted third molar (M3) removal, and to assess the influence of adjunctive guided bone regeneration (GBR).

Methods

This retrospective cohort study included patients presenting with ERR of M2s associated with impacted M3s who underwent M3 extraction with or without concurrent GBR. Cone-beam computed tomography (CBCT) was used to assess ERR progression and distal alveolar bone levels before surgery and at least 6 months postoperatively. Pulp sensibility was also evaluated. Changes in distal bone height were compared between groups. Generalized estimating equations (GEE) were used to evaluate the association between treatment modality and bone gain while adjusting for baseline defect severity and potential confounders. Interaction analysis was performed to assess whether the effect of GBR varied according to baseline defect severity.

Results

The Wilcoxon signed-rank test revealed no significant pre- to post-treatment change in root length among teeth with root length alteration (Z = 0.110, 95% CI − 0.185 to 0.380, P = 0.457), nor in the vertical dimension of the defect among teeth without root length alteration (Z = 0.033, 95% CI − 0.248 to 0.283, P = 0.785), supporting the radiographic stability of the resorptive defect following M3 removal. Regarding pulpal status, excluding one tooth that had undergone preoperative root canal treatment, all remaining teeth (64/64, 100%) maintained positive pulp sensibility at the final follow-up, as assessed by cold testing. The between-group difference was 2.03 mm (Hodges-Lehmann estimate, 95% CI, 0.56–3.82 mm), with significantly greater bone gain in the GBR group (6.32 ± 2.67 mm) compared with the Extraction-Only group (4.08 ± 3.50 mm), P = 0.007. After adjustment for baseline defect severity and other covariates, treatment modality remained a significant predictor of distal bone gain in the GEE model with exchangeable correlation structure (QIC = 1123.229), with GBR associated with greater bone gain compared with extraction alone (β = 2.244, 95% CI, 0.612–3.875, P = 0.007). To assess the robustness of this estimate to potential unmeasured confounding, the E-value was calculated: the E-value for the point estimate was 3.25, while the E-value for the lower bound of the 95% CI was 1.68. The interaction term between treatment modality and baseline distal bone height was not statistically significant (β = 0.008, 95% CI (-0.229, 0.245), P = 0.945), suggesting that the effect of GBR on distal bone gain did not vary significantly across the range of baseline defect severities observed.

Conclusion

Among M2s with available follow-up, no radiographic progression of ERR was detected following M3 removal. The findings do not support routine prophylactic root canal treatment in asymptomatic M2s with preserved pulp sensibility. However, spontaneous periodontal recovery of ERR-associated distal defects appeared incomplete. Adjunctive GBR was associated with enhanced distal bone regeneration.

Clinical relevance

External root resorption of second molars associated with impacted third molars appears to stabilize after extraction of the third molar, with preservation of pulp sensibility. However, the distal bone defect frequently persists radiographically without full resolution. Guided bone regeneration was associated with greater radiographic bone gain in affected second molars.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12903-026-09857-4.

Introduction

The clinical magnitude of external root resorption (ERR) associated with impacted third molars (M3s) has long been obscured by the limitations of conventional radiographic imaging. Historically, the prevalence of ERR was considered negligible, often cited 0.3–13.3% in studies relying on panoramic radiography [1–4]. However, this low incidence appears to be a diagnostic artifact caused by the limitations of 2D imaging. Recent three-dimensional analysis using Cone-beam computed tomography (CBCT) reported incidence rates increasing to 11.0-54.3% [5–9]. This dramatic diagnostic shift underscores that ERR is a widespread, yet frequently underestimated, threat to second molars (M2s).

Removal of the impacted M3 leads to arrest of the resorptive process, and prophylactic root canal treatment may be unnecessary for M2s with ERR. Although Anbu and Song favour prophylactic RCT of M2s with ERR [10, 11], Qu et al. advocate only extracting the M3, reporting that 89% (56/63) of M2s remained asymptomatic with normal responses to cold and heat testing [12].

The distal alveolar bone loss associated with M2 ERR remains insufficiently investigated. ERR is accompanied by distal alveolar bone loss, resulting in a combined defect involving both root structure and periodontal support. The extent to which such defects can undergo spontaneous healing following M3 removal remains insufficiently investigated. Guided bone regeneration (GBR) has been widely used to enhance periodontal bone regeneration, particularly in vertical and contained defects. In the context of M3 surgery, several studies have suggested that GBR may improve distal bone healing of M2s [13–18]. However, existing evidence has largely focused on cases without concomitant root resorption, and it remains unclear whether the presence of ERR alters the regenerative potential or the effectiveness of GBR.

Therefore, the aim of the present study was to evaluate the periodontal healing outcomes of M2s affected by ERR following impacted M3 removal, and to investigate the potential adjunctive benefit of GBR. In addition, the study explored whether the effect of GBR was influenced by baseline defect severity.

Materials & methods

Study design

This single-centre retrospective study was conducted and reported in accordance with the STROBE statement guidelines for cohort studies. The protocol was approved by the Ethics Committee of Peking University Third Hospital (Approval No. IRB00006761-M2024038) and conducted in accordance with the Declaration of Helsinki. Given the retrospective design and use of de-identified clinical and radiographic data, the Ethics Committee waived the requirement for individual informed consent.

Study population

Patients diagnosed with ERR of M2 associated with impacted M3 between 2022 and 2024 were screened. Inclusion criteria were:

1. ERR was defined radiographically as “a clear loss of substance in the root of the adjacent M2 due to direct contact with an impacted third molar” adapted from Al-Khateeb and Bataineh [4] (Fig. 1a).

2. Clinically asymptomatic M2 with normal pulp sensibility. This was strictly defined as a positive response to cold testing (using Endo-Ice) comparable to control teeth, with no lingering pain, and absence of tenderness to percussion or palpation.

3. Extraction of the aetiological M3 performed after the ERR diagnosis.

4. A minimum observation period of 6 months.

5. Availability of high-quality CBCT scans (voxel size ≤ 0.2 mm) pre-operatively and at the final follow-up.

Fig. 1.

Radiographic criteria. a Included case: sagittal CBCT view of a mandibular M2 with external root resorption extending into the root canal system without involvement of the pulp chamber floor (green arrow), associated with an impacted M3. b–d Excluded cases showing caries within the resorptive defect (b, arrow), apical periodontitis (c, arrow), and resorption extending to the pulp chamber floor (d, arrow)

Exclusion criteria included

1. Presence of caries in the resorption area, or apical periodontitis, sinus tract, or mobility ≥ Grade II prior to intervention (Fig. 1b, c). These exclusion criteria targeted pulpal or periapical pathology of non-resorptive origin (e.g., caries-induced pulpitis).

2. Resorption extending to the pulp chamber floor (Fig. 1d).

3. ERR attributed to other aetiologies (e.g., cysts, tumours, or trauma).

4. History of or active orthodontic treatment.

5. Image artifacts (e.g., beam hardening from metal restorations) preventing precise measurement.

Resorption confined to the root canal system, even with frank pulpal exposure, was distinguished from resorption extending to the pulp chamber floor: the former was managed with M2 retention (with GBR considered when pulpal exposure was present), whereas the latter carried a poor prognosis for retention and was therefore excluded from this cohort.

Sample size and grouping

As this was a retrospective study involving consecutive sampling of all eligible M2s within the defined period, an a priori sample size calculation was not performed. Patients were stratified into two groups based on the surgical intervention received: the Extraction-Only Group (M3 extraction alone) and the GBR Group (M3 extraction with concurrent GBR).

Treatment allocation was not randomized. GBR was more likely to be recommended in M2s presenting with one or more of the following conditions: (1) suspected pulpal involvement, defined as radiographic evidence of the resorptive defect confined to the root canal system (Fig. 1a, green arrow) without extension to the pulp chamber floor (Fig. 1d, yellow arrow); (2) radiographic evidence of root length alteration, or (3) a relatively deep distal bone defect (approximately > 8 mm). Final treatment decisions were made jointly by the clinician and patient.

Data collection

Demographic data and clinical records were retrieved from the electronic medical database. The following variables were extracted: gender, age, tooth position, number of affected roots, ERR depth and location, preoperative pulp sensibility status, surgical modality and prognostic outcomes of the M2.

Data extraction was performed by two calibrated examiners. All patient data were anonymised prior to analysis to ensure confidentiality.

Radiographic assessment

CBCT images were acquired using 3D eXam, KaVo, Warthausen, Germany (parameters: 120 kV, 5 mA, FOV 16 cm × 8 cm, voxel size 0.2 mm) and Smart3D, LargeV, Beijing, China (parameters: 100 kV, 6 mA, FOV 12 cm × 8 cm, voxel size 0.1 mm). Despite the use of different devices, image reconstruction and measurement protocols were standardized.

Five specific radiographic parameters were measured on these adjusted planes (Fig. 2):

Fig. 2.

Schematic definition of radiographic indices. a Defect dimensions: root length (r), distance from CEJ to upper margin (u), distance from lower margin to apex (l) and vertical dimension (e). b Distal alveolar bone height (h): CEJ to alveolar crest. c–e Reference standards: root length (r) and homologous root length (c) for cases with compromised CEJ or apex

Root Length (r): The distance from the Cemento-Enamel Junction (CEJ) to the anatomical apex. If the apex was resorbed, measurement was taken to the resorbed tip (Fig. 2c). If the CEJ was compromised, the upper margin of the ERR served as the reference (Fig. 2e).

Upper Boundary Distance (u): The distance from the CEJ to the coronal margin of the ERR defect (Fig. 2a). Recorded as 0 mm if the resorption extended to CEJ.

Lower Boundary Distance (l): The distance from the apical margin of the ERR defect to the root apex (Fig. 2a). Recorded as 0 mm if the apex was resorbed.

Distal Alveolar Bone Height (h): The distance from the distal CEJ to the distal alveolar bone crest (Fig. 2b). A smaller value of h indicates better bone level.

Comparison Root Length (c): The length of a homologous healthy root (e.g., the mesiobuccal root for an affected distobuccal root, or the root of the adjacent first molar if the M2 roots were compromised) (Fig. 2c-d).

Based on these parameters, the Vertical Dimension of ERR (e) was calculated as:

e=r-u-1 (root length didn’t change).

or

e = c-r (root length changed).

Measurements were performed independently by two investigators at both baseline and final follow-up. To assess intra- and inter-examiner reliability, 12 M2s (approximately 18% of the sample) were randomly selected and re-measured by each examiner after a 2-week interval. Due to the radiopacity of the bone substitute materials, the examiners could not be blinded to the treatment group during the postoperative assessment; however, they were blinded to the clinical outcome data. The mean of the two measurements was used for analysis.

Surgical procedures

All surgical procedures were performed by the same experienced oral surgeon. Following local anaesthesia, a full-thickness mucoperiosteal flap was elevated to expose the distal aspect of the M2. The impacted M3 was removed using minimally invasive techniques.

In the GBR group, the extraction socket and distal defect were grafted with deproteinized bovine bone mineral (Bio-Oss, Geistlich, Switzerland), followed by coverage with a resorbable collagen membrane (Bio-Gide, Geistlich, Switzerland). Primary wound closure was achieved in all cases. (Fig. 3)

Fig. 3.

Surgical procedure for impacted M3 removal with guided bone regeneration. a Preoperative CBCT assessment. b–f Key surgical steps, including flap elevation, defect exposure, graft placement, and membrane coverage. g Extracted impacted M3. h Postoperative CBCT at follow-up

All patients received a standardized postoperative regimen regardless of treatment group, consisting of oral cefadroxil (0.5 g twice daily for 6 days), chlorhexidine acetate solution rinse (three times daily for 6 days), and ibuprofen sustained-release capsules (0.3 g as needed for pain control).

Outcome measures

The primary outcome was radiographic bone gain (Δh), defined as the change in distal bone height between baseline and follow-up. Δh was analysed as a continuous variable assuming a normal distribution with an identity link function.

Secondary outcomes included cessation of ERR progression and pulp sensibility status at follow-up.

Statistical analysis

Statistical analyses were performed using SPSS software, version 29.0 (SPSS Inc., Chicago, IL, USA). Because resorptive defects extending to the root apex altered the measured root length, whereas defects confined within the root without apical involvement were bounded by a measurable vertical dimension, pre- to post-treatment stability was assessed using the parameter appropriate to each defect morphology: root length (r) among teeth with apex-involving resorption, and the vertical dimension (e) of the defect among teeth without root length alteration, using the Wilcoxon signed-rank test.

Given repeated tooth-level measurements within patients, generalized estimating equations (GEE) were used to account for clustering of teeth within patients, with the working correlation structure selected as the one yielding the minimum quasi-likelihood under independence model criterion (QIC) among candidate structures. Multivariable models were constructed to evaluate the association between treatment modality and bone gain, adjusting for potential confounders.

In the primary model, the association between treatment modality (GBR vs. Extraction-only) and Δh was evaluated after adjustment for baseline defect severity and potential confounders, including baseline distal bone height, defect ratio, age, and tooth position. The defect ratio was defined as the proportion of the resorptive defect relative to root length; in cases with root length alteration, the relative change in root length compared with the contralateral tooth was used. To assess potential collinearity, Pearson correlation analysis was performed between baseline distal bone height and defect ratio prior to model construction.

To explore whether the effect of GBR varied according to baseline defect severity, an interaction model was constructed, with baseline distal bone height mean-centred to reduce collinearity and an interaction term between treatment modality and centred baseline bone height included. Regression coefficients (β) with robust standard errors and 95% confidence intervals were reported for all effect estimates. To quantify the robustness of the primary GBR effect estimate to potential unmeasured confounding arising from non-random treatment allocation, the E-value was calculated using the approach of VanderWeele and Ding for linear regression estimates. A two-sided P value < 0.05 was considered statistically significant.

Results

Study population and baseline characteristics

A total of 233 M2s were initially screened, of which 65 M2s in 55 patients met the inclusion criteria and were included in the final analysis (Fig. 4). Of the 168 excluded M2s, the reasons were insufficient follow-up (n = 145), caries-induced pulpitis or apical periodontitis (n = 11), no extraction of the impacted M3 (n = 7), and resorption involving the pulp chamber floor of M2 (n = 5). To ensure statistical independence and avoid data clustering within a single tooth, only one root per tooth was included in the quantitative analysis. All included teeth had complete baseline and follow-up radiographic measurements for the primary outcome; no imputation was required. Individual-level measurement data are provided in Supplementary Data 1.

Fig. 4.

Flow-chart of study

Baseline demographic and radiographic characteristics are summarized in Table 1. No statistically significant differences were observed between the GBR group (n = 43) and the extraction-only group (n = 22) in terms of age, sex distribution, tooth position, or baseline radiographic parameters (all P > 0.05) except for a significantly greater baseline distal alveolar bone height in the GBR group (P = 0.043), reflecting more severe disease at baseline in this group. The mean follow-up duration was 7.93 ± 4.67 months (range, 6–28 months) in the GBR group and 9.09 ± 6.53 months (range, 6–28 months) in the extraction-only group. No cases of inferior alveolar nerve injury were observed in either group during the follow-up period.

Table 1.

Demographic and baseline characteristics of the study population

Variables	Total (n = 65)	GBR Group (n = 43)	Extraction-Only Group (n = 22)	P Value	Statistical test
Demographics
Age (years), Mean ± SD	25.77 ± 4.4	26.28 ± 3.16	24.77 ± 3.10	0.091	Independent t-test
Gender (Male/Female), n	22/43	14/29	8/14	0.759	Chi-square test
Tooth Characteristics
Tooth Position (Maxilla/Mandible), n	18/47	11/32	7/15	0.595	Chi-square test
Radiographic Measurements
ERR Position (Cervical/Middle/Apical), n	3/43/19	2/28/13	1/15/6	0.968	Chi-square test
ERR Root Length (r), mm (Mean ± SD)	9.69 ± 2.81	9.53 ± 2.90	10.02 ± 2.67	0.814	Mann-Whitney U test
ERR Vertical Dimension (e), mm (Mean ± SD)	5.25 ± 1.41	5.19 ± 1.40	5.36 ± 1.47	0.658	Independent t-test
Distal Bone Height (h), mm (Mean ± SD)	7.82 ± 3.13	8.30 ± 2.59	6.88 ± 3.90	0.043	Mann-Whitney U test
ERR Severity
Pulp Involvement (Yes/No), n	39/26	25/18	14/8	0.669	Chi-square test
Number of Affected Roots (1 / ≥2), n	63/2	42/1	21/1	0.624	Chi-square test
Root length change (Yes/No), n	19/46	13/30	6/16	0.804	Chi-square test

Reliability of radiographic measurements

Radiographic measurements demonstrated high reproducibility. The intra-examiner ICC was good to excellent (0.662–0.981for measurer 1 and 0.877–0.984 for measurer 2), and the inter-examiner ICC was excellent (0.808–0.957).

Cessation of external root resorption and pulp sensibility

The Wilcoxon signed-rank test revealed no significant pre- to post-treatment change in root length among teeth with root length alteration (Z = 0.745, P = 0.457; Hodges–Lehmann estimate of the median difference, 0.110 mm; 95% CI, − 0.185 to 0.380), nor in the vertical dimension of the defect among teeth without root length alteration (Z = 0.273, P = 0.785; Hodges–Lehmann estimate, 0.033 mm; 95% CI, − 0.248 to 0.283), supporting the radiographic stability of the resorptive defect following M3 removal.

Regarding pulpal status, excluding one tooth that had undergone preoperative root canal treatment, all remaining teeth (64/64, 100%) maintained positive pulp sensibility at the final follow-up, as assessed by cold testing. No signs of periapical pathology or increased tooth mobility were observed clinically or radiographically.

Radiographic distal bone gain outcomes

Concurrent GBR significantly improved radiographic distal bone gain compared to spontaneous healing (Fig. 5). Figure 5e visualizes the comparative recovery of Δh: the between-group difference was 2.03 mm (Hodges-Lehmann estimate, 95% CI, 0.56–3.82 mm), with significantly greater bone gain in the GBR group (6.32 ± 2.67 mm) compared with the Extraction-Only group (4.08 ± 3.50 mm), P = 0.007.

Fig. 5.

Representative cases and quantitative analysis of distal bone regeneration following M3 extraction with or without GBR. a, b Representative case treated with concurrent GBR: preoperative (a) and 6-month postoperative (b) CBCT views demonstrating substantial distal bone regeneration. c, d Representative case treated with extraction alone: preoperative (c) and 6-month postoperative (d) CBCT views showing cessation of ERR with limited bone recovery. e Box plot of distal bone gain (Δh) with individual data points overlaid. The GBR group demonstrated significantly greater bone gain compared with the extraction-only group (Hodges-Lehmann estimate, 2.03 mm; 95% CI, 0.56–3.82 mm; P = 0.007)

To account for potential confounding factors and clustering of teeth within patients, GEE analysis was performed. After adjustment for baseline defect severity and other covariates, treatment modality remained a significant predictor of distal bone gain in the GEE model with exchangeable correlation structure (QIC = 1123.229), with GBR associated with greater bone gain compared with extraction alone (β = 2.244, 95% CI, 0.612–3.875, P = 0.007). To assess the robustness of this estimate to potential unmeasured confounding, the E-value was calculated: the E-value for the point estimate was 3.25, while the E-value for the lower bound of the 95% CI was 1.68.

Baseline distal bone height was also significantly associated with Δh, indicating that sites with more severe initial bone loss exhibited greater absolute bone gain.

Collinearity and interaction analysis

Pearson correlation analysis showed no strong correlation between baseline distal bone height and defect ratio (r = 0.137), indicating that both variables could be included in the multivariable model without significant collinearity.

To explore whether the effect of GBR varied according to baseline defect severity, an interaction model was constructed. The interaction term between treatment modality and baseline distal bone height was not statistically significant (β = 0.008, 95% CI (-0.229, 0.245), P = 0.945), suggesting that the effect of GBR on distal bone gain did not vary significantly across the range of baseline defect severities observed.

Discussion

ERR-associated defects differ from conventional distal periodontal defects following impacted M3 removal. Unlike plaque-driven periodontal breakdown, ERR is initiated by direct mechanical contact between adjacent teeth, resulting in concurrent root substance loss and localized alveolar bone remodelling [19, 20]. This combined involvement of the root surface and supporting bone may influence subsequent healing.

Removal of the causative M3 resulted in cessation of ERR in all cases, with maintained pulp sensibility as assessed by cold testing. These findings are consistent with previous reports and support a conservative approach, as endodontic intervention does not appear to be indicated in the absence of clinical signs of pulpal involvement [12].

Despite increasing recognition of ERR, its periodontal implications remain poorly characterized. Previous studies have focused primarily on the progression of resorption and pulpal outcomes, whereas distal periodontal healing has been investigated mainly in non-resorptive contexts [13, 17, 18, 21]. Whether these findings can be extrapolated to ERR-associated defects remains uncertain.

This gap may be partly explained by the relatively low prevalence of clinically significant ERR, particularly in advanced forms, which limits the feasibility of assembling sufficiently large cohorts. In addition, surgical management of ERR-associated impacted M3s is often more technically demanding. Close proximity and direct contact between teeth may increase the risk of iatrogenic damage to the already compromised M2, which may discourage routine inclusion of such cases in clinical studies, especially those involving regenerative procedures. Furthermore, third molar surgery, while generally safe, carries a small risk of rare but serious complications, such as Lemierre syndrome, underscoring the need for careful case selection [22].

In the present cohort, distal bone defects of varying severity were consistently observed, indicating that resolution of the resorptive process does not equate to complete periodontal recovery. Compared with non-resorptive defects, the altered root surface morphology in ERR may provide a less favourable substrate for spontaneous periodontal reattachment. Resorption-related irregularities of the root surface may compromise clot stability and early wound organization, potentially limiting spontaneous bone fill following extraction alone.

GBR promotes predictable bone regeneration primarily through space maintenance and selective cell exclusion: the barrier membrane physically excludes fast-proliferating epithelial and fibroblastic cells, allowing slower-migrating osteogenic cells to repopulate the defect, while the grafting material provides an osteoconductive—and, depending on its composition, osteoinductive—scaffold that stabilizes the coagulum within the protected space [23–26]. An overview of systematic reviews (191 primary studies) identified ridge preservation as the most effective intervention for distal periodontal healing, yielding a mean 1.21 mm gain in alveolar bone height over spontaneous healing [13], a magnitude broadly comparable to that observed in the present cohort. Age also appears to modulate healing potential: meta-analytic data suggest favourable spontaneous healing in patients under 25 years, whereas older patients more often retain residual defects without intervention [16]. This study demonstrated that GBR yielded 2.24 mm in radiographic distal bone height gain compared to extraction alone, and this association remained significant after adjustment for baseline defect severity and other covariates. No significant interaction between treatment modality and baseline defect height was detected, suggesting that the relative benefit of GBR was observed across the range of defect severities included. However, this finding should be interpreted with caution given the limited sample size and the possibility of insufficient statistical power to detect interaction effects. Notably, patients selected for GBR presented with significantly greater baseline defect severity than those managed with extraction alone, while the GBR group achieved significantly greater bone gain. This pattern runs counter to the direction expected under simple confounding by indication—wherein cases with more severe baseline disease would typically be expected to have a poorer, not better, outcome—and lends indirect support to a genuine treatment effect of GBR beyond what would be predicted by baseline severity alone. Nonetheless, as residual confounding by unmeasured factors cannot be entirely excluded in a non-randomized design, this observation should be interpreted as hypothesis-generating rather than definitive.

Regarding the timing of intervention, our protocol advocates for immediate grafting at the time of extraction. Comparative studies have demonstrated that immediate GBR achieves bone infill comparable to delayed approaches [18, 27]. Taken together, these findings regarding both the efficacy and timing of GBR may warrant consideration of a lower threshold for adjunctive regenerative intervention in ERR-associated cases; however, given the observational design, this recommendation should be validated in prospective studies before clinical adoption.

Limitations

Several limitations should be acknowledged. First, this was a retrospective, non-randomized study conducted at a single centre, in which treatment allocation between GBR and extraction alone was based on clinical judgment rather than randomization. Patients selected for GBR generally presented with more severe baseline defects, and although statistical adjustment was performed for baseline distal alveolar bone height, residual confounding by indication cannot be excluded. A sensitivity analysis using the E-value approach indicated that an unmeasured confounder would need to be associated with both treatment allocation and Δh by a risk ratio of at least 1.68, beyond the measured covariates, to fully explain away the lower bound of the observed association. As this threshold is relatively modest and plausible unmeasured confounders (e.g., smoking status, oral hygiene compliance) could plausibly reach this magnitude, the possibility of residual confounding by indication cannot be excluded, and this finding should be interpreted with corresponding caution. Additionally, the majority of excluded cases (145 of 168) were due to insufficient follow-up, which may have introduced selection bias favouring patients with more regular follow-up behaviour or milder disease, potentially limiting the external validity of these findings.

Second, clinical periodontal parameters such as probing depth and clinical attachment level were not available in this retrospective cohort; radiographic bone gain was therefore used as a surrogate outcome for periodontal healing, and this substitution should be interpreted with appropriate caution. Distal alveolar bone height was quantified using a linear measurement, which offers high reproducibility and straightforward clinical applicability; however, three-dimensional volumetric analysis, which CBCT technically permits, may provide more comprehensive information on bone regeneration and represents a valuable direction for future research. Preoperative smoking status and a comprehensive periodontal assessment could also not be consistently retrieved from the medical records, precluding adjustment for these potential confounders. In addition, data on membrane exposure and graft-related infection were incompletely documented in a subset of cases and could therefore not be reliably analysed as outcomes. Cold testing assesses pulp sensibility rather than true vascular vitality; future studies incorporating electric pulp testing, laser Doppler flowmetry, or pulse oximetry could provide more definitive evidence of pulp health [28–32].

Third, the sample size was relatively small, which may have limited statistical power, particularly for the interaction analysis, for which no adjustment for multiple comparisons was applied given its exploratory nature; a non-significant interaction term should therefore not be interpreted as definitive evidence of a uniform GBR effect across baseline defect severities. Furthermore, the present study evaluated early radiographic healing over a minimum follow-up of 6 months and cannot determine the long-term stability of the observed bone gain.

Finally, this study was conducted at a single centre by a single experienced surgeon in a Chinese population, which may limit the generalizability of these findings to other clinical settings, surgical techniques, or ethnic populations.

Future directions

Prospective, randomized controlled trials with standardized treatment allocation are needed to more definitively establish the efficacy of adjunctive GBR for ERR-associated distal defects and to minimize the confounding by indication inherent to the present retrospective design. Incorporating a more comprehensive outcome assessment—including electric pulp testing alongside cold testing, probing depth, and clinical attachment level—together with three-dimensional volumetric radiographic analysis, would allow more precise and clinically meaningful conclusions regarding both pulpal and periodontal outcomes in future studies.

Beyond the management of established ERR, several upstream questions remain unresolved. Comparative studies of M3 extraction versus retention are needed to clarify whether ERR progresses when the impacted M3 is left in situ, and whether long-term M2 survival differs between these approaches, as some ERR-affected M2s may retain adequate function without intervention. Furthermore, identifying risk factors for ERR—such as M3 angulation and impaction pattern, patient age, and sex—could support risk-stratified models to guide prophylactic M3 extraction before resorption occurs, rather than after ERR has already developed.

Conclusion

Among M2s with available follow-up, no radiographic progression of ERR was detected following M3 removal. The findings do not support routine prophylactic root canal treatment in asymptomatic M2s with preserved pulp sensibility. However, spontaneous periodontal recovery of ERR-associated distal defects appeared incomplete. Adjunctive GBR was associated with enhanced distal bone regeneration.

Abbreviations

M2

second molar

ERR

external root resorption

M3

third molar

GBR

guided bone regeneration

CBCT

Cone-beam computed tomography

CEJ

Cemento-Enamel Junction

GEE

generalized estimating equations

QIC

quasi-likelihood under independence model criterion

Authors' contributions

Baocheng Yao and Shiyu Qiu contributed equally to this work and share first authorship.Baocheng Yao: Conceptualization (Lead); Methodology (Lead); Writing – Review & Editing (Lead). Shiyu Qiu: Data Curation (Lead); Formal Analysis (Lead); Writing – Original Draft (Lead). Chengyi Wang: Investigation (Equal). Qiao Ling: Investigation (Equal). Ning Ma: Supervision (Lead); Validation (Lead). All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by Beijing-Haidian Union Natural Science Fund (Grant No. L222114).

Data availability

Anonymized individual-level measurement data supporting the findings of this study are provided in Supplementary Data 1. Additional data are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The retrospective evaluation was approved by the Ethics Committee of Peking University Third Hospital (Approval No. IRB00006761-M2024038). Given the retrospective design and use of de-identified clinical and radiographic data, the Ethics Committee waived the requirement for individual informed consent to participate.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.