Mandibular Molar Root Resection Versus Implant Therapy: A Retrospective Nonrandomized Study

Molar teeth with furcation involvement represent a treatment challenge that is further complicated by the multitude of available treatment options, which range from scaling and root planning to extraction and replacement with a dental implant.1,^–3 Treatment choice depends on several factors, and degree of furcation involvement is a major determinant not only for the indicated treatment modality but also for prognosis.1^,2^,4,^–7 The long-term prognosis of molars with degree III furcation is poor relative to molars with a lesser degree of furcation involvement.6^,7 

One of the available surgical periodontal treatment modalities for molars with degree III furcation involvement is root resective therapy.8,^–11 The goal of this approach is either to remove single roots with a poor prognosis, for example, because of advanced attachment loss, endodontic treatment complications, root fracture, or subgingival root caries,12,^–14 or to remove plaque niches by transforming multirooted teeth with severe furcation involvement into single-rooted teeth, thus improving access to oral hygiene measures.15^,16 Although several studies have evaluated the outcomes of root resective therapy,11^,17^,18 only a limited number have directly compared root resective therapy with dental implant therapy.18,^–20 

In reviewing the published literature on root resective therapy outcomes, we found that success and complication rates vary depending on the teeth treated; in particular, there appears to be dissociation between maxillary and mandibular molars.18^,21 The same can be said for the success and/or complication rates of dental implants. Similar results have been reported in both rough surface and smooth surface implants.3^,18^,22,^–25 Because of such site-specific differences, in order to be valid, studies must assess and compare therapeutic outcomes at the same anatomic site. Therefore, the purpose of this private practice–based retrospective study was to compare the long-term complication and survival rates of root resected mandibular molars relative to that of dental implants replacing mandibular molars.

Sixty patients (40 men; mean age = 49.9 years, range = 39 to 73 years) treated from January 1993 to December 2001 were included in this retrospective study. All patients received detailed information regarding treatment options and alternatives and were given at least 48 hours to decide on their treatment of choice. Patients were also informed that their clinical pictures, x-rays, and clinical records could be included in future studies submitted for journal publication; those in agreement (their data were used in this study) provided informed consent. All subjects had a history of chronic periodontitis with a minimum of 4 sites with clinical attachment level (CAL) loss > 4 mm, radiographic evidence of alveolar bone loss, and bleeding on probing (BOP) in at least 4 sites.26 Swelling and presence of exudate were used as signs of inflammation. Bony defects were evaluated by the use of radiographs. Chronic periodontitis had been treated previously by scaling and root planning and subsequent surgical periodontal treatment. Prior surgical periodontal treatments were completed at least 6 months before root resective surgery. All patients were treated in a private periodontal practice of one of the authors (G.G.Z.)

The inclusion criteria were as follows : (1) grade III furcation involvement; (2) radiographically estimated residual bone ≥ 50% of the length of the retained root; (3) root resective surgery performed on the first, or the first and second molars only; (4) no existing conditions that might interfere with periodontal or implant treatment; (5) no known drug allergies; and (6) maintenance for at least 48 months. The exclusion criteria were as follows: (1) root resective surgery on second mandibular molars only; (2) implant treatment either in edentulous mandibular molar areas or in the second mandibular molar area only; (3) active periodontal disease; (4) bruxism; (5) smoking more than 10 cigarettes per day; and (6) the presence of pregnancy, diabetes mellitus, history of medication, or drug abuse.

The included patients were placed in either the hemisection-treated group (Group H) or the implant-treated group (Group I). Group H comprised 32 patients with 56 treated molars; Group I comprised 28 patients with 36 implants. The demographics of the study population are summarized in Table 1.

Patients were referred back to their general dentist for endodontic therapy, which in all cases was performed before root resection. Root canals were obturated using a lateral condensation technique and a combination of root canal sealer (AH26, DeTrey Dentsply, Konstanz, Germany) and gutta-percha points. After completion of endodontic treatment, custom-made gold posts (Para Post System, Whaledent, Mahwah, NJ) were inserted and fixed with composite. The coronal portion of hemisected teeth was restored with resin buildup.

All root resection surgeries were performed by the same periodontist. After full-thickness flap elevation, the tooth was sectioned with a diamond burr to separate the 2 roots. The distal half was prepared as a premolar tooth, and the mesial half was extracted. The extraction site was cleaned, filled with a 70:30 composite of bovine-derived xenograft (BDX; BioOss spongiosa 0.25–1 mm, 0.25 g, Geistlich Biomaterials, Wolhusen, Switzerland) and autologous bone, and then covered with a resorbable membrane (Bio-Gide, Geistlich Biomaterials).27 The flap was repositioned and secured with nonresorbable sutures (Ethibond Excel, 3–0, Johnson & Johnson, St-Stevens-Woluwe, Belgium) (Figures 1a through e).

Socket augmentation was performed according to Hoffmann et al.28 An intrasulcular incision extending to the adjacent teeth was made, and a full-thickness flap was elevated. No vertical releasing incisions were made. Following atraumatic extractions, sockets were curetted carefully and irrigated with sterile saline solution. The socket preservation was performed by the use of nonresorbable dPTFE membranes (Cytoplast, Regentex GBR-200, Oraltronics, Bremen, Germany) alone without the use of any soft- or hard-tissue grafts. No further steps were taken to secure the membrane in place. The flap was repositioned and sutured in place with interrupted sutures, and the membrane was then removed under local anesthesia at week 4 (Figures 2a through c).

Extraction, socket preparation, and preservation also followed the aforementioned treatment protocol. Cylindrical screw-type implants (Straumann, RN, 4.1 mm, SLA, Institut Straumann, Waldenburg, Switzerland) were placed 8 months after extraction, using a one-stage surgical approach. Following full-thickness flap elevation, osteotomy preparation was performed at 875 rpm, and implants were placed manually, at a torque of 35 Ncm (046.119/046.049 Institut Straumann) (Figure 2 d).

Patients in both groups were prescribed a systemic antibiotic (clindamycin 600mg, Ratiopharm, Ulm-Donautal, Germany; 600 mg/day for 4 days) to be taken once a day for 6 days and the oral analgesic diclofenac (Voltaren 100 mg, Novartis Pharma, Nuernberg, Germany; 100 mg/day for 4 days), starting 1 day before surgery. Patients were also instructed to rinse twice daily with 0.1% chlorhexidine (Chlorhexamed Fluid, GlaxoSmithKline, Buehl, Germany) for 3 weeks, also starting 1 day before surgery. Sutures were removed 8 to 10 days after surgery. Follow-up appointments were scheduled twice a month during the first 2 months after the extraction surgery and once a month for the following 10 months, after which patients were assigned to routine maintenance care.

All prosthetic restorations were made by the same restorative dentist. Hemisected first molars were restored with temporary fixed partial denture (FPD) immediately after surgery (Figure 1f) and with a metal-ceramic FPD 6 months latter. FPD included the second premolar as an abutment. When both molars underwent hemisection, the FPD included the distal half of the second molar, the distal half of the first molar, and the second premolar (Figures 1g and h). All restoration margins were placed subgingivally. Final implant restorations were delivered 6 months after implant placement (Figures 2e and f).

Complications were divided into 7 types. Periodontal pocket formation and osseous defects with CAL > 5 mm and signs of inflammation were classified as type 1 complications; CAL was defined as the distance between crown margin and deepest probing point (rounded to the nearest millimeter) and recorded with a periodontal probe (UNC 15, Hu-Friedy, Leimen, Germany). Additional classification was as follows: root caries or caries at the crown margin (type 2); apical abscesses (type 3); root fractures (type 4); peri-implantitis with an augmentable osseous defect (≤ 50% of the implant length; type 5); peri-implantitis with a nonaugmentable osseous defect (> 50% of the implant length; type 6); loss of the implant before final prosthesis loading (type 7); and finally, prosthetic complication, including fracture, screw loosening, and so on (type 8).

For the treatment of type 1 and 5 complications, a flap was elevated, and the root or implant was debrided. The periodontal or peri-implant osseous defect was grafted with a 70:30 mixture of BDX and autologous bone. In patients experiencing type 2, 3, 4, and 6 complications, the failing hemisected molar(s) or implant(s) were extracted or explanted; extraction/explantation sites were preserved as previously described.28 

All patients were enrolled in a maintenance program consisting of semiannual follow-up appointments where supra- and subgingival debridement was performed, teeth (or implant restorations) were polished, and oral hygiene instructions were given. The first maintenance visit was performed at the time of final prosthesis delivery. CAL, BOP, and plaque index (PLI)29 using disclosing solution (Mira-2-Ton, Hager & Werken, Duisburg, Germany) on 4 surfaces of each tooth (remaining root) or implant supporting the fixed partial dentures were recorded. Measurements at the deepest point of the selected defect are reported herein. Baseline CAL, BOP, and PLI values were collected at the first follow-up visit and compared to respective values obtained at the end of the observation period. CAL changes were analyzed as absolute CAL loss (in millimeters), percentage of CAL loss from baseline, and annual rate of CAL loss as percentage loss from baseline per year.

Descriptive statistics of CAL, BOP and PLI, as well as the incidence of complications and time in maintenance are presented. Nominal scaled values are reported as absolute and percentage frequencies; interval scaled values are reported as means and ranges when normally distributed and as medians and ranges when non-normally distributed, as determined by the Kolmogorov-Smirnov test. Parametric and nonparametric methods were used respectively to analyze statistical differences for normally and non-normally distributed interval scaled values.

Age, gender and smoking habits were evaluated for possible influence on complication rates in each group. Independent-samples t test was used to analyze age, and Fisher's exact test to analyze gender and smoking habits.

In the case of interval scaled values, differences between the 2 time points (baseline and end of observation period) were tested with the Wilcoxon matched-pairs test; differences between the 2 treatments were tested with the Mann-Whitney U test. Nominal scaled values were compared using McNemar's test in the case of matched samples (frequencies at different time points) and using χ2 test in the case of independent samples (frequencies in different groups). The χ2 test or Fisher's exact test was applied to test differences in incidence of complications. The time in maintenance or until the first occurrence of a complication was demonstrated by a Kaplan-Meier curve; differences were examined with the log rank test.

For the purpose of statistical analysis, each treated tooth or implant was considered a unit. A significance level of 5% was applied. P values were adjusted for multiple comparisons within the same data set. ³⁰ For statistical evaluations, a commercially available software was used throughout (SPSS for Windows 15/2006, SPSS, Chicago lll).

As shown in Table 2, complications occurred in 18 teeth (32.1%) in Group H; 6 (33.3%) diagnosed as salvageable and 12 (66.7%) as nonsalvageable. Nonsalvageable complications included root caries in 6 cases (50.0%), apical abscess in 2 (16.7%) and root fracture in 4 cases (33.3%). Meanwhile, among the 36 implants of Group I, 4 (11.1%) experienced complications; 3 of which (75.0%) were salvageable and 1 (25.0%) was nonsalvageable. There were no instances of implant loss before prosthetic loading.

The overall risk of occurrence of complications was greater in Group H than in Group I (χ2 test, P = .021 / P_adj = .042; odds ratio = 3.79) (Table 2). Although the incidence risk for nonsalvageable complications was almost 10 times greater in Group H than in Group I (χ2 test, P = 0.010/P_adj = .030; odds ratio 10.11) (Table 2), salvageable complications did not differ significantly between the groups (Fisher's exact test, P = .482). A wide time range (2 to 65 months) in the posttreatment maintenance duration before the onset of complications was present (Table 2). No significant correlations between complication rate and age, gender, or smoking habits were found for either group (P > .1). For Group I, no correlation was found between the incidence of salvageable or nonsalvageable complications and implant dimensions (data not shown).

Kaplan-Meier analysis (Figure 3) of the maintenance duration indicated a mean complication-free time of 71.4 months for Group H and 74.3 months for Group I compared with 85.0 months for Group H and 76.4 months for Group I for salvageable complications. The mean time to incidence of a nonsalvageable complication was 77.7 months for Group H and 77.8 months for Group I. A group comparison revealed group effects on the incidence of complications (P = .027) and the incidence of nonsalvageable complications (P = .013). There was no group effect on the incidence of salvageable complications (log rank test, P = .619).

BOP at baseline was present in 46% and 22% of cases in Groups H and I, respectively, compared with 29% and 39% of cases, respectively, at the end of observation period. There were no significant intergroup or intragroup differences (for either time point) in BOP frequency (P > .075). Similarly, no significant intragroup or intergroup differences were noted in PLI frequencies at baseline (39% and 28% for Groups H and I, respectively) or at the end of the observation period (21% and 39% for Groups H and I, respectively). Neither baseline BOP and PLI nor final PLI values correlated with the occurrence of complications for either group (P > .05). A relationship between final BOP and complication rates was observed for Group H (Fisher's exact test, P = .004/P_adj = .016) but not for Group I (P > .05).

The CAL data are presented in Table 3. There were significant intragroup differences between baseline and final values for both groups (Wilcoxon test; P = .001/P_adj = .007 for Group H; P < .0001/P_adj < .0016 for Group I). There were also significant intergroup differences for CAL at both time points (U test, P < .0001/P_adj = .0016). However, the absolute and percentage CAL loss, as well as the percentage of CAL loss per year, did not differ significantly between the 2 groups (U test, P_adj > .05).

In Group H and Group I there were significant differences in CAL loss per year between cases with and without complications (U test; P = .002/P_adj = .012, Group H; P < .0001/P_adj = .0016, Group I) (Table 3). Additionally, there were significant differences in Group H between salvageable and nonsalvageable subgroups for all CAL loss measurements (Table 3).

This retrospective study compared the complication/survival rates of root resected mandibular molars with that of dental implants replacing mandibular molar teeth, during a maintenance care period of 4 years. The results indicated that both root resected mandibular molars and mandibular molar implants could be expected to have, on average, a complication-free survival of 6 years. Although root resected molars were at a significantly greater risk for complications, approximately 80% of root resected mandibular molars were retained overall, and almost 70% of root resected mandibular molars remained complication free for an average of 5 years.

Our findings in implant cases were similar to previously reported ranges for implants in mandibular molar positions. Survival rates of 98.4% have been documented in retrospective analysis.24 Cornelini et al31 showed a 1-year survival rate of 96.7% for implants replacing mandibular molars supporting single crowns, and a 1-year success rate of 97.5% for mandibular molar implants supporting FPDs.32 Success and survival rates of implants do, however, decrease with longer follow-up periods. Becker et al23 reported a 6-year cumulative success rate of 91.5% for implants in mandibular molar positions, while cumulative 10-year survival rates of 93%33 and 93.7%.34 have been reported for mandibular implants. In a long-term analysis of molar implants (follow-up from 6 to 125 months), noted complications included suppuration in approximately 11.1% of vertical implants and 2.5% of pockets; 88% of the implants in the current study were used to replace mandibular molars.3 Success rates of 98.6% and 85.0% for implants inserted in the first and second mandibular molar position, respectively, over a 15-year period, were reported by Fugazzotto,18 with the majority of failures observed during the first 7 years, using the criteria outlined by Albrektsson.³⁵

According to the Albrektsson criteria, bone loss exceeding minimal amounts is regarded as “failure.” However, such criteria do not allow for an estimation of the degree of destruction. For clinical purposes, it is of interest whether an implant can be salvaged or needs to be removed. To our knowledge, no published data exist distinguishing “salvageable” from “nonsalvageable” implant complications; consequently, the authors' clinical experience was used determine this end point. The latter demonstrated that treatment of implants with bone loss exceeding 50% of the implant length does not lead to satisfactory long-term results. Therefore, these implants were removed.

The prognosis of dental implants placed where teeth were lost due to periodontal disease was previously shown to be worse than for implants placed as a result of tooth loss for other reasons.36 In the same study, implant complications were limited to peri-implantitis, which could be managed in 3 of 4 cases,36 similar to the findings of the present study.

Although several studies of root resective periodontal therapy have provided information about failure and complication rates, few provide relevant data specific for mandibular molars. Hamp et al37 reported various periodontal procedures, including 44 mandibular root resections, most of which involved amputation of the mesial root, similar to our findings. No losses were reported up to 5 years. According to Langer et al,16 36% of root-resected mandibular molars failed over 10 years, mostly due to root fracture, periapical abscesses, and caries.16 Of 366 mandibular molars resected by Fugazzotto,18 11 were classified as failures over a 15-year period; there were 5 failures in the first 7 years, 4 between years 7 and 9, and 2 between years 9 and 15, for an overall 96.8% success rate. Eight of these failures were in cases where the distal roots were amputated. Suemnig et al38 reported 38 resected mandibular molars (35 mesial root amputations) with 5 failures/extractions noted during the 5- to 6-year observation period, consistent with the hemisection therapy results of the present study.

Discrepancies in the reported long-term success rate of root-resected molars may stem from the fact that a multitude of factors may influence treatment outcome. Among these are tooth type, parafunctional habits, endodontic treatment protocol, and prosthetic treatment protocol. In the present study, only mandibular first and second molars were included, and inclusion was further limited to molars in which the mesial root was removed to eliminate additional anatomic variables. Fugazzotto18 found that 30% of root resection failures and 38% of molar implant failures were associated with untreated parafunction, which has been suggested as a potential risk factor for late complications in dental implants.39^,40 Patients with bruxism or parafunctions were excluded from the present study.

It has been suggested that the reason for the low endodontic success rate in some root resection studies is the lack of a controlled endodontic and prosthetic protocol.41 However, the rate of endodontic complications was low in the present study, despite the lack of a standardized endodontic protocol (endodontic therapy was provided by the general dentists); a possible explanation for this finding is case selection and the relatively short follow-up time. The lack of prosthodontic failures in our study was consistent with the reported low incidence of such complications in root resected molars,12 and it might be attributed to the chosen fixed prosthesis design18 and the relatively small number of cases. Although smoking was a significant risk factor for the long-term prognosis of both dental implant43^,44 and periodontal surgical therapy,43 smoking was not associated with complications in the present study, a finding likely due to the exclusion of patients who smoked more than 10 cigarettes per day. We found that root caries accounted for half of the nonsalvageable complications, suggesting that more effective anticaries measures could reduce the rate of failures among hemisected molars. Given that approximately 80% of hemisected mandibular molars were retained during the observation period, and the large number of factors implicated in determining the outcome of root-resection therapy, case selection becomes critical for this treatment approach. Each case must be carefully evaluated to assess whether adequate endodontic, prosthodontic, and maintenance therapy is feasible, including considerations related to surgical access (root anatomy) and patient motivation.

In the present study, hemisected mandibular molars and mandibular molar implants experiencing complications had a greater annual rate of relative CAL loss than teeth and implants that remained complication free. In the hemisected teeth, this result can be explained by the CAL-based definition of periodontal complication. Meanwhile, in the case of implants, previous studies have shown a strong correlation between implant CAL loss and bone loss,45 which was used to define implant complications.

No type 8 complications (prosthetic complications, screw fracture or loosening) were identified in the current study. Complications were further classified as salvageable (types 1 and 5) and nonsalvageable (types 2, 3, 4, 6, and 7) teeth or implants. To our knowledge, no published data exist define the difference between “salvageable” and “nonsalvageable” implant complications; consequently, the authors' clinical experience was used. In our experience, treatment of implants with bone loss exceeding 50% of the implant length does not lead to satisfactory long-term results. Therefore, these implants were removed. No subcategory of complication was created for cases of suppuration as our study did not observe suppuration in the implant site in the absence of bone loss.

Increased values for BOP and PLI did not necessarily indicate an insufficient oral hygiene to justify exclusion of these patients from the study. The indices used were taken into account. A major disadvantage of the plaque index used lies in the fact that after discoloration any visible staining is recorded as a positive count irrespective of the amount of plaque present. Therefore even small amounts of plaque lead to a positive count. The use of a more differentiated index would have been better, but it was not possible because of the limitations of patient treatment in a private practice setting.

In summary, within the limitations of this retrospective, practice-based study, implants replacing periodontally involved mandibular molars had fewer complications than hemisected mandibular teeth. Subsequent prospective clinical trials are needed to confirm and allow generalization of these findings.

The authors report no conflicts of interest related to this study. No financial or material support was provided by any company to the authors or the patients involved in this study. The authors thank Mrs. Ulrike Schulz, statistical consultant, for the statistical analysis of the data.