The limit of tolerable micromotion for implant osseointegration: a systematic review

Much research effort is being invested into the development of porous biomaterials that enhance implant osseointegration. Large micromotions at the bone-implant interface impair this osseointegration process, resulting in fibrous capsule formation and implant loosening. This systematic review compiled all the in vivo evidence available to establish if there is a universal limit of tolerable micromotion for implant osseointegration. The protocol was registered with the International Prospective Register for Systematic Reviews (ID: CRD42020196686). Pubmed, Scopus and Web of Knowledge databases were searched for studies containing terms relating to micromotion and osseointegration. The mean value of micromotion for implants that osseointegrated was 32% of the mean value for those that did not (112 ± 176 µm versus 349 ± 231 µm, p < 0.001). However, there was a large overlap in the data ranges with no universal limit apparent. Rather, many factors were found to combine to affect the overall outcome including loading time, the type of implant and the material being used. The tables provided in this review summarise these factors and will aid investigators in identifying the most relevant micromotion values for their biomaterial and implant development research.

Micromotion and osseointegration. One human and twenty-four animal studies were identified (Table 1). For the human post-mortem study, the micromotion for osseointegrated hip stems was less than 40 µm which compared to 150 µm for a stem with failed bone ingrowth. For the animal studies, the mean value of micromotion for implants that osseointegrated was 32% of the mean value for those that did not (112 ± 176 µm osseointegrated versus 349 ± 231 µm non-osseointegrated, Mann Whitney test p < 0.001, Fig. 2). However, the osseointegration outcome also depended on other experimental/implant conditions with no distinct osseointegration limit detected. Rather, the range for successful/failed osseointegration overlapped: 15 to 750 µm for osseointegrated samples versus 30 to 750 µm for non-osseointegrated samples (Table 1 and Fig. 2).
The effects of research method: applied vs measured micromotion. When micromotion was applied, lower micromotion resulted in more consistent osseointegration (Fig. 3A, Mann Whitney p value = 0.001). Similarly, when micromotion was measured at the end of the study duration, implants that osseointegrated had lower micromotion than implants that did not (Fig. 3B, Mann Whitney p value = 0.01). Comparing values of micromotion between the methods (measured vs applied), no differences were observed for the osseointegrated group, and similarly there was no difference between the methods for the non-osseointegrated group. (Fig. 3). Records  Observation time and Bone-implant-contact. There was a positive correlation between observation time and percentage BIC, with longer study duration time resulting in better percentage BIC (Spearman's ρ = 0.40, p value = 0.01).

Discussion
The most important finding of this systematic review was that the available data refutes the idea of a universal limit of tolerable micromotion for implant osseointegration. Whilst on average, the micromotion associated with osseointegration was 32% of the micromotion associated with failed fixation, many exceptions to the rule were identified (Fig. 2). In some studies, micromotion at the bone-implant interface as high as 750 µm osseointegrated, whilst in other micromotion as low as 30 µm did not osseointegrate. Thus, implant and external factors must be considered when estimating the level of micromotion that could lead to successful osseointegration for a new biomaterial/implant. The following implant factors were associated with higher levels of micromotion and successful osseointegration: hydroxyapatite coating 43 , larger threads in lower density bone 8 , and square pore cross-sectional shape 37 . The following environmental factors were associated with higher levels of micromotion and successful osseointegration: infrequent loading 31 , a rest period following initial loading 41 and longer study duration (9 weeks or more) 30,44 (Table 2). The gold standard micromotion limit is often considered 150 µm. However, Overgaard et al., and Soballe et al., showed that this level of micromotion can be tolerated if a period of rest is allowed after initial loading and if the implant was coated with hydroxyapatite 41,43 . The accelerated resorption of HA coating under excessive micromotion could have led to a better bony ingrowth as studies have previously shown that HA coating on Table 1. Osseointegrated (OI) and non-osseointegrated (Non-OI) values of micromotion (µm) from the studies selected For applied values, the value was set as a controlled experimental parameter, for measured values means and standard deviation are reported where possible. *0 represents experiments with immobilized implants after a period of loading. *12 represents experiments that applied an additional implant displacement for 12 weeks.

Author
Year Country Micromotion OI (µm) www.nature.com/scientificreports/ titanium implants improve BIC through its direct interaction with osteoblast, osteoclasts and pro-inflammatory markers 52,53 . Further, previous studies have also shown that a period of rest or otherwise referred to as recovery phase may be beneficial for BIC. The recovery phase or time off helps to counteract the waning effects of long-term mechanical loading, and improve the responsiveness of osteoblasts and osteocytes to restart bone formation 54  www.nature.com/scientificreports/ the pore-cross sectional shape of the outer bone chamber from round to square in the traditional bone chamber designs, bony ingrowth would be facilitated even with micromotions as high as 500 µm.
Whilst the data reviewed revealed that the mean value of micromotion for successful osseointegration was 112 µm, some studies showed that micromotion as low as 30 µm can lead to failed implant fixation 30,40 . The authors attributed this to the duration of loading, hypothesizing that the process of bone formation had not been reached within 6 weeks. Subsequent experiments measured osseointegration at 9 weeks or more and demonstrated successful osseointegration with the longer study duration 30,44 . The data from this systematic review further supports this finding, demonstrating a positive correlation between osseointegration (measured by percentage BIC) and study duration (Fig. 4). Biologically, osseointegration starts with woven bone formation, followed a period of remodelling to lamellar bone in response to mechanical loading. This transition from woven bone to lamellar bone formation takes starts around 6-8 weeks and can take a period of months to complete 56,57 . Therefore, in vivo experimental studies exploring osseointegration of implants should allow a time period of over 6 weeks to see the full healing response. The effects of study duration have also been reflected by recent computational research which highlighted the differing mechanisms between bone healing and remodelling, and hence the importance of the measurement time point 2 .
Another explanation for the contradictory in vivo data is that micromotion is a simplified, clinically convenient measure, which overlooks the fundamental mechanobiological mechanisms that drive implant osseointegration. In vivo data coupled with finite element analyses suggest that it is the interfacial stress-strain state resulting from implant micromotion that stimulates osseointegration 58,59 . Different loading conditions (axial, shear, torsional, etc.), combined with different localised implant/bone geometry lead to different stress-strain states, with too much strain leading to fibrous tissue formation [58][59][60][61] . Indeed, by considering the interfacial stress-strain state, it is possible to relate implant bony ingrowth theory 2,58,59 to fracture healing theory [62][63][64][65] , which intuitively one would expect given the involvement of the same cell types 66 . Conversely, when the implant/environmental conditions that affect the interfacial stress-strain state are ignored, counter-intuitive trends can be observed. For example, when neglecting these factors, it was found that increased micromotion was positively correlated with increased percentage BIC (Fig. 5). However, within study data demonstrate that micromotion and percentage BIC are negatively correlated 50 . This further emphasises the needs to consider implant and environmental factors and their link to the interfacial stress-strain state when interpreting how micromotion affects osseointegration. It should also be noted that there is no standard interpretation of BIC and so caution should also be applied when interpreting BIC between studies. Some studies report BIC as the fraction of mineralized bone in direct contact with the implant surface 44 , whilst other describe it as the length of the implant surface in contact with (both mineralised and non-mineralised) bone relative to the total implant length 34 .
Historically, the causal effect of implant micromotion on osseointegration was investigated by applying known displacement and subsequently measuring osseointegration. However, more recently, research method has shifted to applying known loads and quantifying micromotion at the end of the experiment 4,8 . Both measurement techniques were able to identify differences in micromotion between implants that osseointegrated and those that did not. Interestingly, when comparing the results between the two different methods, no differences were found.
The majority of studies identified applied micromotion as a controlled experimental condition, meaning that mean and standard deviation micromotion data were not available prohibiting application of the established meta-analysis approaches recommended by Borenstein et al 67 . For the same reason, it was not possible to perform   www.nature.com/scientificreports/ a cumulative meta-analysis to quantify the risk of bias between studies. Rather, to provide some quantitative analysis we applied Mann Whitney tests to the data extracted from each study. Further we isolated the effects of studies that applied micromotion, and those that measured it, and found that this did not affect the principal finding our systematic review (Fig. 3).
In conclusion, this systematic review has demonstrated that the idea of a universal limit of tolerable micromotion for implant osseointegration is misleading. Rather, implant and environmental factors, and their link to interfacial stress-strain states, must be considered to identify the most appropriate limit for the biomaterial/ patient group under consideration. The tables provided in this systematic review summarise the implant and environmental conditions for all published quantitative in vivo micromotion research and will enable investigators to compare their data to the most appropriate values.

Materials and methods
Protocol and registration. Prior to the investigation, the protocol was registered with the International prospective register for systematic reviews PROSPERO (ID: CRD42020196686), following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and checklist 68 . Information sources and search strategy. An electronic search was performed for articles published up to 16 th November 2020, in the following databases: PubMed, Scopus and Web of Science. The search strategy identified papers which included the following terms: (micromotion OR "micro-motion" OR "micro motion") AND ("osseointegration" OR "osteointegration").
Study selection. Two independent reviewers (N.K., J.S.) assessed the titles and abstracts of all the studies and discarded studies that met any of the exclusion criteria. The full text of all remaining studies was then assessed against the inclusion and exclusion criteria. Any disagreement regarding eligibility of articles were resolved by a third reviewer (R.v.A.).
Data collection process and data items. Data relating to osseointegrated and non-osseointegrated values of micromotion were extracted. The country, animal species, number of study groups, duration of the experiment, implant material and loading conditions were recorded. The outcome of osseointegration measured as bony ingrowth or percentage bone-implant-contact (BIC) were also recorded. The micromotion methodology (applied or measured) was also recorded. In the applied group, known values of micromotion in the form of cyclic loading were directly applied as a controlled experimental condition and then osseointegration was assessed. In measured group, micromotion was not a controlled experimental condition, rather micromotion at the bone-implant interface was measured once the implant had osseointegrated/not. Statistical analysis. Data were analysed and plotted using Graph Pad Prism 8 software and have been reported as mean ± standard deviation (SD). Four analyses were performed: (1) All micromotion values were grouped into osseointegrated/not, according to the definition used by the original study authors. Data were first tested for normality, and then non-parametric Mann Whitney tests were used to compare differences between groups. (2) Micromotion values were further discretised according to the study method (applied vs measured micromotion). Then analysis 1) was repeated for both of these subgroups. (3) Spearman correlation tests were used to examine correlation between percentage BIC and micromotion values for three groups: all data, osseointegrated, non-osseointegrated. (4) Spearman correlation tests were used to examine the correlation between percentage BIC and study duration.
The significance level was set to α = 0.05.

Data availability
Data generated and analysed during this study are included in this published article. Data are available from the corresponding author subject to reasonable request.