Marginal bone loss around oral implants supporting fixed versus removable prostheses: a systematic review

Aim The aim of this systematic review was to evaluate and compare the marginal bone loss (MBL) around implants of fixed (FISP) versus removable implant-supported prosthesis (RISP). Material and methods This review was conducted according to the PRISMA guidelines. A systematic search of the literature on Web of Science and Ovid (MEDLINE) was conducted in March 2019 to identify randomized controlled trials/quasi-randomized trials, prospective and retrospective studies written in German and English. Two reviewers screened the identified papers for eligibility and performed an independent data extraction. The Newcastle-Ottawa Scale was used to evaluate the level of evidence of the included studies. Results The search resulted in 2577 studies, of which 42 were selected for full-text evaluation. Finally, six studies were included in qualitative analyses, reporting results from 248 participants (81 FISP versus 167 RISP). Five of the included studies were prospective and one study was retrospective. MBL was highest in the first year after implant placement and ranged from 0.17 ± 0.07 mm to 2.1 ± 1.6 mm in FISP and from 0.22 ± 0.55 mm to 2.5 ± 2.7 mm in RISP. After 4 years, there was no statistically significant difference between the groups; MBL ranged from 0.36 ± 0.22 mm to 1.5 mm in FISP and 0.56 ± 0.45 mm to 1.4 mm in RISP. Of the six included studies, two each were rated as good quality, fair quality, and poor quality. Conclusion Fixed and removable implant-supported prostheses seem to have similar long-term outcomes regarding marginal bone loss. However, the evidence provided in this systematic review is limited due to the poor quality of two of the included studies. Future studies with study designs specified to the topic of this review are necessary to provide clear information about marginal bone level alterations in modern implant therapy.


Introduction
Since the investigation of the osseointegration of oral implants by Brånemark in the 1950s and 1960s, implantsupported restorations became a widespread alternative to removable prostheses [1,2]. Both then and now, edentulism is a serious health problem involving functional, aesthetic, phonetic, and psychological problems [3,4]. Although the Fifth German Oral Health Study illustrated a decrease in edentulism among 65 to 74-year olds from 24.8% in 1997 to 12.4% in 2015, therapy concepts are still needed which focus on the resulting disabilities [5]. As the epidemiological findings indicate a decrease in edentulism, treatment concepts for partial edentulism and single-tooth replacement re also considered important in order to meet patient requirements.
The success of implant rehabilitation relies on the integration of the implants in hard and soft tissues. Marginal bone loss (MBL) is, therefore, a critical factor affecting the clinical outcome [6,7]. While moderate MBL of < 0.2 mm per year are generally accepted as within the limits of a normal physiological process, excessive MBL, in particular in the first year after implant insertion, is associated with an increased risk of periimplantitis and tissue collapse affecting not only the survival rates of oral implants but also aesthetics especially in the anterior visible zone [8]. Multifactorial reasons for early excessive MBL are assumed; however, they are not fully understood. The main theories have been the infection theory, supported mainly by periodontists, and the overload theory supported by prosthodontists/restorative dentists [9]. However, there is clear evidence that combined factors affect MBL and a single-minded explanatory model for bone loss is not acceptable [9,10]. In a systematic review, Qian et al. found no evidence that overloading alone represents the incriminating factor behind marginal bone resorption around oral implants [9]. Furthermore, aseptic foreign body reactions and strong immune and inflammatory reactions of the individual host immune system play a key role in MBL progression [9,11,12]. MBL primarily occurs in the early stage after implant placement. While an early MBL of 1.0-1.5 mm during the first year has long been assumed as normal, newer data shows losses of 0.459-0.55 mm in the first year after implant insertion [7,[13][14][15][16]. Surgical trauma, occlusal overload, peri-implantitis, microgap, biologic width, and implant crest module have been reported to be positive causative factors for early implant crestal bone loss [17]. Albrektsson et al. and Smith and Zarb proposed bone loss of less than 0.2 mm annually after the implant's first year of service as one criterion for implant success [7,14]. In a recent study, an early MBL of > 0.44 mm after the first 6 months of prosthetic loading was suggested as a risk factor for the progression of peri-implant bone loss [18].
For implant-supported restorations, the fixed screwretained and cemented as well as removable restorations are established treatment options [19,20]. For removable implant-supported prostheses, the number of supporting implants, prosthesis design, and the attachment type are important factors for outcome analysis [21,22]. Survival rates for removable or fixed implant prostheses were found to be similar when patients were matched for bone quantity and quality, implant number, implant length, and splinting [19,23].
To the authors' knowledge, the long-term effect on the marginal bone level around implants of implantsupported fixed and removable restorations has not been compared and systematically evaluated in a review. Thus, the aim of this systematic review of the available literature was to evaluate the effect of the restoration type (fixed or removable) on the marginal bone level of oral implants.

Material and methods
This review was conducted based on the guidelines of the transparent reporting of systematic reviews and metaanalysis (PRISMA) and in accordance with the Cochrane handbook for systematic reviews of interventions [24,25].

Search strategy
The research question in the PICO format was formulated as follows [26] Free-text terms were used to search the Web of Science database (Table 1). Medline via OVID was searched with a combined medical subject heading and a free-text term search strategy (Table 2). Then, text-wordtruncation was applied to retrieve all forms of the search terms and Boolean logical operators were used to combine the search results.

Eligibility criteria
The evaluation of titles and abstracts as well as full-text analysis was conducted according to the eligibility criteria listed in Table 3.

Study selection and data extraction
The selection of studies was a two-step process. In the first step, two reviewers (B.S. and M.P.) independently screened the titles and abstracts to determine which Table 1 Electronic free-text term search in Web of Science #1 TS=(((impl* OR overdentur* OR maxillofacial OR "maxillo-facial") NEAR/5 (prosthes* OR fixed OR permanent OR removable OR restorat* OR reconstruct*))) #2 TS=(((bone* OR osteo*) NEAR/5 (alveolar OR maxill* OR mandib* OR jaw OR bucc* OR "peri-implant*" OR periimplant* OR marginal) NEAR/5 (loss OR resorption OR defect* OR density OR atroph* OR distraction OR lyses OR osteolys* OR osteonecro* OR necrosis OR volume))) #3 TS=((predict* OR score* OR observe*) OR (trial OR random* OR therapeutic*)) #4 #2 AND #1 #5 #4 AND #3 studies met the inclusion criteria. In the second step, they assessed the eligibility of the selected studies based on a full-text analysis. While screening the publications, the following parameters were extracted into an extraction chart: author, complications of the inserted implants and restorations, mean age of groups, mean MBL, observation period, number of patients participating, number of patients in the groups, number of implants, study design, dental status, survival and success rate, year of publication. Search agreement between the two reviewers was evaluated with the Cohen's Kappa (k) test. Any disagreements were resolved by a discussion with a third reviewer.

Risk of bias
The risk of bias and quality analyses of the studies were performed by the two reviewers independently. The Newcastle-Ottawa Scale (NOS) was used to analyze the included studies [28]. Studies can have score values between 0 and 1 in the "selection" and "exposure/outcome" categories and between 0 and 2 in "comparability". As there are four items in the "selection" category, one item in the "comparability" category and three items in the "exposure/outcome" category, the maximum score according to the NOS is nine. A score of "0" was awarded when the criterion was not satisfied, a score of "1" was awarded when the criterion was satisfied, and for "comparability", a score of "2" was awarded when the criterion was satisfied using a validated method.
The NOS score was used to categorize overall study quality as either good quality (> 7), fair quality (6-7), or poor quality (< 6).

Statistical analysis
Descriptive analyses were performed to calculate weighted means and standard deviations (SD) using standard spreadsheet software (Excel for Mac 2011, version 14.5.2, Microsoft, Redmond, WA, USA).

Screening process and study characteristics
The search resulted in 1083 studies from Medline and 2024 studies from Web of Science ( Fig. 1). A total of 530 were duplicates. After scrutinizing the titles and abstracts, 2535 studies were excluded due to not meeting the inclusion criteria, leaving 42 publications for full-text analyses. After examining the full texts, 36 papers were excluded because they did not meet the defined inclusion criteria ( Table 4). The remaining six studies were included in the systematic review. Five of the included studies were prospective and one was retrospective. The number of patients ranged from 30 [29] to 74 [30] with a mean of 43.7 ± 14.7. The number of investigated oral implants ranged from 60 [29] to 235 [30] with a mean of 160.2 ± 56.9, supporting a mean of 20.2 ± 13.8 FISP and 21.2 ± 7 RISP. The number of implants inserted per patient ranged from 5 to 6 for FISP and from 2 to 4 for  RISP. One study [31] did not report the overall number of implants and three studies [29][30][31] did not report the number of implants inserted per patient for the FISP and RISP group ( Table 5). The k values for potential article inclusion (titles and abstracts) and selected studies were 0.79 and 0.88, respectively, indicating excellent agreement between the reviewers.

Assessments of quality and risk of bias
Two studies [30,34] were rated as overall poor quality, mainly because of their potential risk of bias regarding the comparability domain. Two studies [31,33] were considered fair quality and two [29,32] as good quality (Fig. 2).

Data synthesis
The observation period of the selected studies ranged from 1 to 10 years with a mean of 4.8 ± 2.7 years. MBL measurements were evaluated using detailed narrow beam radiography (DNB) [29,32] or long cone intraoral radiography [30,31,33,34]. In four studies [30][31][32][33], the region of implant-abutment connection was used as the baseline reference for the MBL measurements in mm. Measurements were taken mesially and distally to determine bone level changes. In one study [29], the authors used a scoring system using the threads of the implant as a measuring scale for mesial and distal measurements, providing no mm values. However, the data was used for comparability analyses between FISP and RISP regarding MBL in the given observation period. One study [34] did  not provide information about the measurement technique. MBL was most significant in the first year after implant placement, ranging from 0.17 ± 0.07 mm to 2.1 ± 1.6 mm for FISP and 0.22 ± 0.55 mm to 2.5 ± 2.7 mm for RISP (Table 6). Palmqvist et al. used a scoring system, thus conversion to mm was not possible. Weighted arithmetic means of the score were calculated for each of the given years and listed in Table 6.
In all studies, the data suggested a similar outcome between FISP and RISP regarding MBL at the end of the observation period. According to the available data, MBL seems to initially be higher in RISP but does not differ between the groups after 4 years. Two studies [33,34] observed initially higher MBL in FISP. However, this observation was either not significant for all observation periods [34] or no information was provided regarding statistical significance for any of the given observation  Different attachment types were used in the RISP groups among the included studies. The overdenture (RISP) patient population examined by Quirynen et al. [31] was divided into three groups: a "bar" group, provided with an egg-shaped Dolder bar; a "magnet" group  with two open-field magnets; and a "ball" group with two ball attachments. Data on MBL was only provided overall, however, not for each separate group. Palmqvist et al. provided no data regarding attachment types, as the overdenture group (RISP) consisted of a subpopulation of consecutive patients from an earlier study [29,35]

Discussion
Despite great achievements in global oral health, edentulism remains a major and irreversible problem affecting quality of life. In providing adequate therapy options for these patients, information regarding outcomes plays a key role in decision-making. Thus, the aim of the present review was to systematically evaluate the differences in MBL-an important factor for the long-term success of oral implants-between FISP and RISP [6,7]. The included studies were published in peer-reviewed scientific journals and judged to be reliable. Nevertheless, an assessment of the potential risk of bias was performed using the Newcastle-Ottawa Scale to indicate the level of evidence for the main research question. For long-term observations, the available data suggest that both therapy options result in a similar amount of MBL. Reported MBL after 1 year of loading was 2 mm to 3.2 mm for RISP (depending on the attachment type and implant number) and 1.5 mm for screw-retained FISP [23,36,37]. The difference in MBL between screwretained and cemented FISP was evaluated in several studies. While Nissan et al. [38], Lemos et al. [39], and Hameed et al. [40] reported greater MBL in screwretained FISP, Koller et al. [41] and Sailer et al. [42] found that cemented FISP showed more MBL (> 2 mm) than screw-retained restorations. In contrast to these findings, systematic reviews by Brandao et al. [43] and Sherif et al. [44] did not find evidence for a significant difference in MBL between screw-retained and cemented restorations. After 1 year of observation, MBLs reported in the included studies ranged from 0.17 ± 0.07 mm to 2.1 ± 1.6 mm in FISP and 0.22 ± 0.55 mm to 2.5 ± 2.7 mm in RISP. Other authors reported similar findings regarding MBL values after 1 year of observation ranging from 0.1 mm to 1.5 mm [13,14,45]. In all included studies providing data from more than 4 years, MBL for RISP and FISP was similar after 4 years [29,[31][32][33]. Only one included study [30] investigated MBL in both jaws. The values described in mm in four of the included studies focused on osseointegrated implants in the mandible [31][32][33][34]. As described in Table 3, Palmqvist et al. used a score system; thus, conversion to mm was not possible. Cune et al. reported MBLs of 2.6 ± 2.1 mm for the maxilla and 1.9 ± 1.5 mm for the mandible. In contrast to previous findings of Quirynen et al. [37], where most subjects with maxillary overdentures showed continuous bone losses of up to 3 mm or more after 53 months of loading, data provided by Palmqvist et al. do not support the theory of continuous bone loss in patients provided with overdentures [29,37]. However, Palmqvist et al. reported that the poorest bone scores were found in three overdenture patients [29].
This review included data for up to 10 years; thus, the long-term effect of the two different therapy options could be examined. Furthermore, three of the included studies, reporting data from more than 4 years for 100 patients overall, were rated as "good quality". This underlines the strength of the evidence for the conclusions of this review.
Nevertheless, there are several limitations to this review that need to be addressed. As described by Palmqvist et al., different results for MBL outcomes are likely due to patient selection criteria, length of the placed implants, design of the prostheses, and different operators with different surgical techniques [29]. The previously mentioned factors limit the comparability of different studies. The prostheses' attachment type is an important confounding variable and is crucial to take into account in interpreting the results. Due to the different attachment types used across the studies, comparisons are limited. Two of the included studies [32,34] used one attachment type for all overdenture patients. The other studies either did not provide enough information regarding the attachment type [29,33] or used various attachment types in their overdenture group [30,31]. No studies provided data on MBL between different attachment types, and therefore, the effect of attachment type on MBL remains unanswered. Furthermore, the timing of the baseline measurements and a description of the measurement technique are important to consider in interpreting the findings. Nevertheless, these crucial details are often not well described. In the present review, two studies were rated of "poor quality" mainly due to a lack of comparability between the intervention groups. in these studies' primary research questions. While Palmqvist et al.'s focus was on marginal bone level alterations in the intervention groups, MBL was not the primary outcome in the other included studies. The above-mentioned limitations and the low number of relevant studies emphasize the need for betterdesigned studies of the differences in MBL between FISP and RISP. The authors encourage future investigators to conduct prospective studies according to the guidelines described by the Cochrane handbook for systematic reviews of interventions, with control patients matched for factors affecting MBL outcome [24].