A systematic review and meta-analysis of the survival rate of implants placed in previously failed sites

The aim of this study was to conduct a systematic review and meta-analysis to assess the clinical outcomes of dental implants placed in previously early and late implant failed sites. An electronic literature search was conducted in several databases for articles published up to February 2018. Human clinical trials that received at least one implant in a previously failed site were included. Hence, the PICO question that was aimed to be addressed was: Do patients undergoing implant replacement (second and third attempts) in previous failed sites have survival rates similar to implants placed at first attempts? A random effects model was used to calculate survival weighted means and corresponding 95% Confidence Intervals (CI) among studies. Eleven studies of low to moderate methodological quality were included in this review. Implants placed in sites with history of one and two implant failures had a weighted survival rate (SR) of 88.7% (95%CI 81.7–93.3) and 67.1% (95%CI 51.1–79.9), respectively. Implants placed in sites with a previous early failure revealed a weighted SR of 91.8% (95%CI 85.1–95.6). First implants presented higher SR than implants placed in sites with one or two previous implant failures. In contrast, implants placed in sites with one and two implant failures had similar SR. Within its limitations, this review suggests that replacement implants have moderate SR. Larger prospective studies with well-defined criteria for early and late implant failure are necessary to confirm and expand on these results.


Introduction
Although osseointegrated implants are routinely used for the rehabilitation of partially or totally edentulous patients, presenting high long-term survival rates; 1,2,3,4 biological and technical complications may result in implant failure and loss. 5Implant failures have been reported in frequencies varying from 1% up to 22%. 1,2,3,4Factors affecting implant failure are diverse and are related to patient systemic status, age and social habits, implant macro-/micro-design and surface chemical composition, implant position, bone quality, and surgical technique. 6,7,8Implant failures have been classified as early or late depending on its time of occurrence. 5arly failures occurring before or at abutment connection, as an inability to establish intimate bone-to-implant connection, 4 and late failures occurring after prosthetic loading and related mainly to plaque-induced peri-implantitis and, possibly, to occlusal overloading. 9n some cases, implant replacement is the only treatment that allows fixed rehabilitation in sites with previous implant failure. 10In other instances, it is considered the therapy of choice owing to superior occlusal stability, 11,12 masticatory efficiency and personal self-esteem provided by dental implants, 13,14,15,16,17 without damaging adjacent teeth.Nonetheless, implant replacement still represents a challenging scenario. 18lthough a number of retrospective cohort studies 18,19,20,21,10,22,23,24,25,26 and one prospective clinical trial 27 have evaluated survival rates of implants placed at sites with history of implant failure, these studies included limited numbers of patients.As such, the precise prognosis of implants placed at failed sites remains unclear.A recent meta-analysis conducted by Quaranta et al. 28 showed that the survival rates of implants placed in the sites of previously failed implants are low.However, the specific prognosis of implants placed in sites with history of early and late implant failed remains to be determined.Therefore, the main aim of this study was to conduct a systematic review and meta-analysis to evaluate the survival of dental implants placed in previously early and late implant failed sites.The following focused question was addressed: "Do patients undergoing implant replacement (second and third attempts) in previous failed sites have survival rates similar to implants placed at first attempts?

Methodology Protocol and registration
The study protocol of this systematic review and meta-analysis was registered at the National Institute for Health Research PROSPERO, International Prospective Register of Systematic Reviews (registration number 51293).The text was structured in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines. 29

Type of studies and participants
Controlled clinical trials, retrospective studies and prospective case series that received at least one implant in a previously failed site were considered eligible for inclusion.

Exclusion criteria
Studies that included patients who had received replacement implants in sites other than the failed ones were excluded.Animal studies and case reports were also excluded.

Intervention and comparison
Survival rates of implants placed in sites with early and late implant failure.

Outcome measures
The primary outcome was the survival rate of implants placed in previously early and late failed sites.

Information sources and search strategy
Search strategies were developed for MEDLINE, Embase, Lilacs, SciELO, Scopus, Web of Science databases, as well as the grey literature.Medical subject headings (MeSH) terms were combined with Boolean operators and used to search the databases.All searches were performed without language restriction, up to February 2018.The following keywords and MeSH terms were used: ((("dental implantation, endosseous"[mh] OR "dental implants"[mh])) AND ((((replantation [mh]) or (failed implant [tiab]) or (implant replacement [tiab]) NOT ("animals"[mh]).The electronic search was complemented by manual searches of reference lists from selected full-articles.

Study selection
Two calibrated reviewers (CCV and CF) screened independently titles and abstracts.Studies meeting the inclusion criteria, or those with insufficient information in the title and abstract to make a clear decision, were selected for further evaluation.Disagreements were solved by a discussion between the reviewers.All studies that met the inclusion criteria underwent a validity assessment.

Data collection, risk of bias assessment in individual studies
Data collection and quality assessment of included studies were performed independently by two reviewers (CCV and CF), using the modified Newcastle-Ottawa Scales (NOS). 30Disagreements were solved by discussion with a third reviewer (CMP).

Summary measures and synthesis of results
Inter-investigator reliability was assessed using kappa coefficient.Data from all studies examining the survival rate of implants placed in sites with history of one and two implant failures were pooled and analyzed using statistical analysis software (Comprehensive Meta-Analysis software, v.3.3.070,Englewood, USA).§ A random effects model was used to calculate the weighted means and corresponding 95% Confidence Intervals (CI) among the studies.Statistical heterogeneity and consistency among the studies was assessed with I. 2

Study characteristics
Characteristics of the final 11 studies are reported in Table 1.Outcomes of implants placed at sites with history of one implant failure were reported in ten studies (Table 2), while outcomes of sites with two implant failures were reported by five studies (Table 3).

Implants placed at sites with history of one implant failure
Studies provided data on 579 patients treated with 704 replacement implants (Table 2).Time interval between removal of failed implants and placement of the replacement implant varied significantly, ranging from 0 (immediate replacement) to 132.8 months (Table 2).Replacement implants were followed for less than 1 year in two studies, 23,25 and for at least 1 year in seven studies. 18,19,20,21,10,27,24In of one study, the observation period of replacement implants was not reported 26 (Table 2).
Out of the 704 replacement implants placed in sites with history of one implant failure, 94 failed (Table 2), leading to a weighted survival rate of 88.7% for these implants (95%CI: 81.7-93.3;I 2 = 78%, p = 0.000) (Figure 2A).According to the studies that reported the time of the replacement implant failure, most occurred within the first months of follow up, but ranged from 1 to 158.9 months (Table 2).Second failure reasons and types were reported solely in two studies 20,10 and most were classified as late.The causes for replacement implants loss were lack of osseointegration, peri-implantitis, occlusal overload or unknown causes (Table 2).
Only three studies reported on success rates of implants placed in sites with history of one implant failure. 20,27,25The random effects model revealed a weighted success rate of 85.0% for    these implants (95%CI: 66.3-94.2;I 2 = 67.85%,p = 0.045) (Figure 2B).Success criteria, however, was not disclosed in one study, 25 and the other two studies adopted distinct success classifications: Wang et al. 20 considered success as absence of pain, mobility, history of exudates, and bone loss around implants not exceeding 2 mm; while Quaranta et al. 27 used the criteria proposed by Albrektsson et al. 31 Studies reporting exclusively on implants placed in sites with history of one early implant failure revealed a weighted implants survival rate of 91.8% (95%CI: 85.1-95.6;I 2 = 48,80%, p = 0,14) (Figure 2C). 20,27,26Implant success rate in these studies varied from 71 to 93.75% (Table 2). 20,27None of the studies reported the outcomes of implants placed exclusively in sites with history of late implant failure.
A weighted comparative analysis of the survival rate of replacement implants placed immediately after removal of failed implants and that of delayed replacement implants placed after healing of failed sites was not performed, as this comparison was addressed by only one study, 23 in which survival rates of immediate and delayed replacement implants did not differ statistically (90.9% and 86.8%, respectively; p = 0.433).

Implants placed at sites with history of two implant failures
Five studies reported outcomes of 45 replacement implants placed at sites with history of two implant failures (Table 3).Time of implant replacement was reported in only two studies (Table 3). 18,22These last implants were followed up to 185.5 months (Table 3).
Out of the 45 implants placed at sites with history of two implant failures, 13 failed (Table 3).The random effects model revealed a weighted survival rate of 67.1% (95%CI: 51.1-79.9;I2= 0.00%, p = 0.47) (Figure 2D).Only one study reported on the type of the failures; 18 and implant success rates were not reported.

Patient characteristics
Patient characteristics varied considerably among the studies.Several studies excluded patients with a) systemic conditions likely to affect bone metabolism, 19,20,10,22,23,24 b) changes in environmental conditions known to affect the outcome of implant therapy (smoking cessation or onset), 24 c ) behavioral conditions (poor oral hygiene, drug abuse habits), 27 d) parafunctional habits, 24,27 and e) heavy smoking habits. 20,23ccording to results from other studies, no correlations were found between replacement implant failure and a) smoking, 20,10,22,24 and b) American Society of Anesthesiologists (ASA) Physical Status classification system. 10However, a higher rate of replacement implant failure was described for patients taking antidepressants and antithrombotic agents. 18oreover, local bone characteristics such as poor quality and low quantity have also been associated with higher replacement implant failure. 18

Replacement implants characteristics
Surface of replacement implants were reported in six studies, analyzing medium rough, 18,19,22 SLA, 20,27 TiUnite and machined surfaces. 18,26Only two studies compared the survival rates of smooth-or rough-surface replacement dental implants, 18,26 demonstrating that the survival rate of rough replacement implants was significantly higher than that of machined implants.

Comparison of the survival rates of implants placed in sites with or without history of implant failure
Due to the large heterogeneity among studies, comparisons of the survival rates of implants placed in sites with history of none, one or two implant failures included only same-study comparative data.Only three studies disclosed the survival rates of implants placed in sites of pristine bone (first implant) and with history of one implant failure. 18,20,25According to a metaanalysis of these studies, first implants had higher survival rates than implants placed in sites with one previous implant failure (96% and 79%, respectively; odds ratio (OR) = 5.19 (3.77-7.16).A lack of a sufficient number of studies precluded a comparison between the survival rates of first implants and implants placed at a site with history of one early failure.
A meta-analysis of the data from two studies that compared survival rates of first implants and implants placed in sites with history of two implant failures 18,25 revealed that the survival rate of first implants was higher than that of implants placed in sites with two previous failures (93.6% and 62.5%, respectively; OR = 8.77 (3.18-24.23).Finally, a metaanalysis comparing the survival rates of implants placed in sites with history of one and two implant failures 18,10,23,25 revealed similar survival rates (82.5% and 76.7%, respectively; OR = 1.52(0.61-3.78).

Methodological quality of included studies
Included studies received a 5-to 9-point score, with a mean of 5,91 points, and thus, were of low to moderate methodological quality due to their risk of bias (Figure 3).None of these studies reported on sample size calculations, training/calibration of examiners, management of confounders or independent blind assessment.Patients included in three studies 18,25,26 were truly representative of patients that received dental implants in sites with previous implant failure in their centers.Patients included in other three studies 19,20,23 were partially representative of patients treated in their center.Other studies failed to describe the derivation of the groups.
In all but one study, ascertainment of sites treated with dental implants was done using secure patient records.Quaranta et al. 27 conducted clinical and radiographic examinations.Definitions of inclusion and exclusion criteria were clearly stated in seven studies. 10,18,19,20,22,24,27efinitions and assessment of implant failure were clearly reported in five studies. 10,20,22,24,27Finally, adequacy of follow-up of patients was only found in two studies. 20,27

Discussion
Evidence based dentistry has had an increasing impact on oral care over the last years.Along these lines, this systematic review and meta-analysis provide a comprehensive assessment of the survival and success rates of implants inserted in sites with history of implant failure(s).We found that the survival rates of first implants ranged from 93% to 99% 18,20,25 and were in line to the data reported in the literature. 1,2,3,4Further, first implants presented survival rates higher than implants placed in sites with history of one or two implant failures.Moreover, implants placed in sites with one and two previous implant failures had similar survival rates.
Although some studies reported the time and causes of implant failure, most have failed to run separate survival analysis for replacement implants placed in sites with history of early and late implant failures.The three studies reporting exclusively on implants placed in sites with history of one early implant failure 20,26,27 revealed a weighted implants survival rate of 91.8%.However, no study reported on the outcomes of implants placed exclusively in sites with history of late failure.Finally, all studies failed to provide survival rates for replacement implants placed specifically in sites with history of biological, mechanical or iatrogenic implant failures.It is critical to note that the pathogeneses of the various types of implant failures are quite distinct. 8,32As such, distinct types of implant failures may differentially affect replacement implant outcomes.
Both patient-and implant-related factors can profoundly affect both the survival and clinical success of osseointegrated implants.Patients with systemic and environmental conditions known to compromise implant therapy outcomes were excluded from many of the studies included in this meta-analysis. 10,19,20,22,23,24,27Exclusion of these patients precluded a more thorough evaluation of the risk factors associated with replacement implant failure and possibly resulted in higher replacement implant survival rates.Thus, the results from this metaanalysis must be interpreted with caution for patients with known risk factors.
A higher failure rate of implants placed in sites with history of implant failure was described for patients taking antidepressants and antithrombotic agents and in sites with poor bone quality and low quantity. 18Selective serotonin reuptake inhibitors (SSRI), a much common prescribed antidepressant, affects not only nervous tissue but may cause bone loss by inhibiting bone-remodeling process triggered by mechanical loading. 33Moreover, its use has already been associated with increased failure risk of dental implants placed in pristine bone (hazard ratio, 6.28). 34oncerning implant-related factors, two studies demonstrated treated-surface implants presented significantly higher survival rates than smooth machined implants. 18,26This result is in agreement with others that showed that rough-surface implants have better survival rates than machined implants placed in pristine bone. 35ri-implantitis is the major condition associated with late implant failure.Recent evidence demonstrated that peri-implantitis progresses in a non-linear accelerating pattern and that its onset occurs following three years of implant function. 36Thus, it is important to highlight that most included studies followed replacement implants for a period of one year or less.Hence, longer follow-ups are necessary to fully disclosure the incidence of late failures in replacement implants.Finally, the included studies are of low to moderate quality level and must be confirmed by high-quality studies.

Conclusions
Despite the low to moderate quality level of the published evidence.existing data indicate that replacement implants have moderate survival rates and can be considered as an approach in the treatment of edentulous patients.Larger prospective studies with well defined criteria for early and late implant failure, larger sample sizes and adequate follow up are necessary to confirm and expand on these results.

Figure 3 . 30 *
Figure 3. Methodological quality of included studies.Study quality was categorized as high (11 to 14 total star points), moderate (8 to 10 stars), or low (0 to 7 stars), according to a modified Newcastle-Ottawa Scales (NOS).30

Table 1 .
Data of implants placed at sites with no history of failure.PCS = prospective case series; RCC= retrospective cohort study.Reported data limited to early failures only; +Data limited to failed implants that were replaced by another implant; ♦ Data limited to failed implants that were replaced twice by implants.

Table 2 .
Data of implants placed at sites with history of one single implant failure.

Table 3 .
Data of implants placed at sites with history of two implant failures.
Figure 2. Survival and success rates of replacement implants.A) Weighted survival rate of implants placed at sites with history of one implant failure.B) Weighted success rate of implants placed at sites with history of one implant failure.C) Weighted survival rate of implants placed at sites with history of one early implant failure.D) Weighted survival rate of implants placed at sites with history of two implant failures.