Introduction: To evaluate the survival and success rate of premolars with a previous root canal treatment which underwent to non-surgical retreatment or to extraction and implant replacement.

Materials and methods: Patients visiting in the Clinical Endodontic Section of the School of Dentistry (University of Bologna) were eligible for the study. Out of these patients, those with previously root canal treated premolars fulfilled the inclusion criteria. Teeth were considered retreatable and restorable (Endo group) or not retreatable and not restorable (Implant group) on the basis of clinical and radiographical examination. Each patient was included in an annual recall programme and final evaluation was performed after 8 years. Clinical parameters were recorded. Differences in baseline characteristics between the two treatment groups were assessed using logistic regression analysis with clustered standard errors. Results were expressed for each variable as odds ratios (ORs) of implant rehabilitation to root canal retreatment with 95% confidence intervals (CIs). Survival and treatment success for the two study groups were estimated using the Kaplan-Meier method. The association of treatment group with time to event was assessed using Cox proportional hazard regression analysis with clustered standard errors to allow for intragroup correlation within teeth belonging to the same patient. Results were expressed as hazard ratios (HRs) of experiencing the study event among implants as compared to root canal retreatments with 95% CIs.

This historical cohort study evaluated the survival and success rate of first and second premolars with a previous root canal treatment which underwent secondary root canal treatment vs. extraction and implant replacement.

The study was planned as a historical cohort of patients followed for 8 years. This work was written according to Consolidated Standards of Reporting trials guidelines for reporting clinical trials, and was conducted in full accordance with ethical principles, including the Declaration of Helsinki. At the time of treatment, an informed consent was signed by each participant.

Patients treated at our university department from January 2007 to December 2015 were eligible to be included in the study.

The final evaluation was performed between January 2019 and March 2023 in the University Department.

The primary consideration for inclusion in the study was that both approaches had a sound clinical and biological rationale and were acceptable in terms of cost for the patient. Nevertheless, the final decision was always made by a pool of operators based on their experience and adherence to evidence-based dentistry and best clinical practices.

The following parameters were considered for the decision-making process of each treatment:

• Structural parameters: residual coronal structure (3 or 4 walls, 1 or 2 walls, 0 walls, full prosthetic reconstruction); presence of deep carious lesions (yes, no).
• Periodontal parameters: distance between cervical enamel junction and first bone contact (CEJ-MBL) (≤3 mm, > 3 to < 6 mm, ≥6 mm) (Endodontic parameters: initial PAI (1 or 2, 3 to 5); preoperative post (yes, no).

Non-surgical root canal retreatment was performed in accordance to previously-published clinical protocols. Detailed information is reported elsewhere. All procedures were made by a pool of tutors of the endodontic master programme.

In all cases, local anaesthesia and dental dam isolation were performed. A straight-line access was prepared using a diamond burs mounted on high-speed water-cooled handpieces (W&H, Burmoos Austria). Gutta-percha solvents were used to soften the obturation core and then an initial pathway was created with Gates-Glidden burs #3 - #4 (Dentsply Maillefer, Ballaigues, Swiss) to approximately 5-6 mm depth in the gutta-percha. After the complete removal of the obturation materials, a manual k file crown-down instrumentation technique was performed. Working length was established with the aid of an electronic apex locator (Root ZX, Morita, Tokyo, Japan) and radiographically confirmed.

The irrigation protocol included a total amount of 3.0 mL of 5% NaOCl. In case of root calcification, each canal was subjected to further irrigation with 3 min of 1.0 mL 17% EDTA solution.

A warm carrier-based technique (Thermafill, Dentsply DeTrey, Konstanz, Germany) with AH Plus sealer (AH Plus, Dentsply Dental Trey, Konstanz, Germany) was used to complete the root canal filling. Filling was repeated immediately when short fillings (<2 mm from the radiological apex) or presence of voids/unfilled canals was identified on periapical X-rays.

All teeth were restored within 2 to 3 weeks following root obturation. An adhesive 2-mm thick build-up was applied using a light-cured glass-ionomer cement (Vitrebond, 3M, St.Paul, MN, USA) followed by the application of phosphoric acid etchant solution for 20-30 s and a bonding system (Clearfil New Bond or Clearfil Photo Bond, Kuraray Co., Ltd, Osaka, Japan) applied according to manufacturer's directions and total-etch technique. Photo-curable composite materials (Clearfil Photo Posterior, Kuraray Co., Ltd, Kyoto, Japan; Z100, 3M) were applied and photo-cured using a multilayering technique to complete the restoration. Prosthetic rehabilitations, usually performed 3 to 6 months after root canal retreatment were performed in a high percentage of cases (46 out of 68 [68%]). All teeth were occlusally loaded and were not used as abutments for multiple fixed restorations.

Extraction and implant insertion procedures were made according to previously published clinical protocols.

Different implant insertion timings were performed according to the tooth periapical pathology and presence of periapical infection. Immediate implant placement was made when no periapical infection was present at the moment of extraction.

For immediate placement, a 1.2 mm drill was used to prepare the alveolar socket, following the palatal bony walls as a guide. A series of calibrated drills at 225 rpm were used under copious irrigation with sterile saline solution. Primary implant stability was obtained by anchoring the implant in the remaining apical portion of the socket at least 3 mm beyond the root apex area. Titanium implants were placed to keep the blasted surface at cortical bone level smooth portion of the neck at soft tissue level. A 1.0 mm cover screw was then positioned and maintained for all the healing phases.

For early placement, a period of approximately 3 or 6 months was waited. A flapless or a flapped technique was performed. A 1.2 mm drill was used to mark the position, angle and depth. The drill passed through the mucosa (transmucosal), cortical bone and cancellous bone under copious saline irrigation. A series of twist and calibrated drill at 225 rpm was used and a site of the adequate depth and diameter was created whilst irrigating with sterile saline solution. A cover screw of 1 mm was then positioned and maintained for all the healing phases.

Loading procedures were performed 3 months after implant insertion. Briefly, impressions were taken in customised resin trays with polyether materials (Permadyne and Garant, 3M ESPE, St Paul, MN, USA). Customised abutments were delivered after 7-15 days and provisional crowns cemented in the same session. After one month, definitive metal-ceramic rehabilitations were positioned and fixed with a polycarboxylate powder/liquid cement (Heraeus Kulzer GmbH, Hanau, Germany).

The periapical radiographs and clinical data were used to classify the final outcome in both treatments. Each patient was checked during the routine recall visit and inspected by one of two examiners for coronal/crown integrity, periapical radiographic status (Endo group) and radiographic bone level stability (Impla group). Radiographs were taken using the paralleling technique and dental intraoral films (Kodak, Rochester, NY, USA). Exposure time of each filling was standardised and a film holder was used (Rinn Corp., Elgin, IL, USA). Radiographic evaluation was performed pre- and post-operatively, every 1-2 years, when the clinical symptoms or coronal status required a further radiographic inspection, and at the end-point by one additional examiner blinded to the study. The following periapical X-rays were analysed by two independent examiners: before the treatment, at the moment of the treatment (root canal filling or implant insertion), at 4-year and at 8-year follow-up.

Prior to radiographic evaluation, the examiners were calibrated using well-defined instructions and reference radiographs with various types of periapical lesions. At the end-point, periapical tissues were classified on the basis of the Periapical Index score (PAI) as follows:

• Healthy: absence of radiographic signs of periapical lesions (PAI ≤2), and absence of clinical signs and symptoms including no tenderness to percussion. Healthy teeth determined the success rate of the study.
• Endodontic lesion: radiological signs of endodontic disease (PAI ≥3) during follow-up.
• Survival rate: number of healthy and endodontic lesion still functional at the end line of the study.

Complications were assessed and divided as:

• Minor: complications that did not affect the endodontic retreatment outcome and survival (e.g. prosthetic complications).
• Major: complications that affected the endodontic outcome/healing and required a reintervention (e.g. persistent periapical lesion).
• Fatal: complications that affected the survival rate of the tooth (e.g. fractures).

The crestal marginal bone and the bone-implant contact were examined to evaluate the marginal bone level (MBL). MBL was assessed at the mesial and distal implant surfaces by measuring the distance between the reference point of the implant platform to the most coronal bone-to-implant contact level using a scale divided into 0.1 mm steps and corrected according to the known height and width of each implant. Radiographic evaluation was performed in single-blind by two additional examiners. Before evaluating the radiographs, the examiners were calibrated by using well-defined instructions and reference radiographs with different marginal bone level measures.

Survival rate was calculated as the number of implants still functional at the end line of the study. Success rate was calculated according to traditionally accepted criteria and included lack of mobility, lack of infection or suppuration, MBL <2.0 mm in the first year, and <0.2 mm at subsequent years .

Complications were assessed and divided as:

• Minor: complications that did not affect the implant outcome and survival (e.g. prosthetic complications);
• Major: complications that affected the implant success and required a reintervention (e.g. severe MBL losses);
• Fatal: complications that affected the survival rate implant (e.g. peri-implantitis).

Numerical variables were summarised as mean ± standard deviation; categorical variables were summarised as frequencies and percentages. Crude differences in baseline characteristics between the two treatment groups were assessed by means of simple logistic regression analysis with clustered standard errors to allow for intragroup correlation within teeth belonging to the same patient. Results were expressed for each variable as odds ratios (ORs) of implant rehabilitation to root canal retreatment with 95% confidence intervals (CIs). A similar approach was used to compare mean decision-making scores, but a linear model was used in place of a logistic model.

Survival and treatment success for the two study groups were estimated using the Kaplan-Meier method using the date of surgery as the time origin and treating losses to follow-up as right-censored data. The association of treatment group with time to event was assessed using Cox proportional hazard regression analysis with clustered standard errors to allow for intragroup correlation within teeth belonging to the same patient. Results were expressed as hazard ratios (HRs) of experiencing the study event among implants as compared to root canal retreatments with 95% CIs. The proportional-hazards assumption was confirmed after checking for nonzero slope of scaled Schoenfeld residuals on time.

In a sensitivity analysis, Cox regression was rerun using weights based on propensity-for-treatment scores in order to fully balance baseline demographic and anatomical characteristics in the two study groups. Multiple Additive Regression Trees (MART) gradient boosting was used to estimate the propensity to get one treatment or the other with a set of explanatory variables including sex, age, tooth location, and tooth type. As a rule of thumb, the following settings were adopted for regularisation: maximum tree depth of 5 interactions; maximum of 20,000 iterations; 50% bagging; 0.01 shrinkage factor. Each observation was weighted by the reciprocal of the probability of receiving the treatment that was actually received, which is known as inverse probability treatment weighting (IPTW). Weights were truncated at the 99th percentile.

Mean differences in PAI (for Endo group) and MBL (for Implant group) were estimated with linear regression analysis with clustered standard errors.

All analyses were carried out using Stata software, version 17. The significance level was set at 5%, and all tests were two-sided.

Patients were considered eligible for the clinical protocol and recall programme if they fulfilled the following inclusion criteria:

• Voluntary healthy patients;
• Age 18-75 years;
• Local geographic provenience;
• Similar socio-economic condition;
• Presence of at least one premolar with a previous root canal treatment requiring a nonsurgical retreatment or extraction and implant rehabilitation;
• Possibility to attend to an annual hygienic recall.

Exclusion criteria were the following:

• General contraindication to implant surgeries;
• Absence of implant restoration or endodontic treatment;
• ASA score >2;
• Diabetes or any condition that could compromise bone healing or immune response;
• Corticosteroid treatment, pregnancy or breast feeding;
• Heavy smoking (>10 cigarettes/day);
• Exposure to radiation therapy focused on the head and neck region;
• Malignant disease directly involving the jaws.

Voluntary healthy patients needing a root canal retreatment or an extraction and implant placement