The effect of non-steroidal anti-inflammatory drugs on the osteogenic activity in osseointegration: a systematic review

The systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [15]. The review is registered at PROSPERO database, and the review protocol can be accessed at http://​www.​crd.​york.​ac.​uk/​PROSPERO/​display_​record.​php?​ID=​CRD42016051448.

The PROSPERO registration number is CRD42016051448.

The review included in vitro, clinical and in vivo studies; animal models. Articles published since the introduction of selective COX-2 inhibitors in 1999 were included [3]. Studies published outside this time period, not in the English language, non-peer reviewed, and review studies were excluded.

An electronic search into four databases: Ovid, Pubmed, Scopus, and Web of Science was performed to systematically identify the available literature. Articles published between January 1, 1999, and July 7, 2018, were considered.

The focus question, used to guide the search strategy, according to the PICO schema is “Will variables such as the dosage, duration of administration, and selectivity of post-operative NSAIDs impair the bone healing around titanium implants?”

The search string comprised the combination of medical subject headings (MeSH) and free keywords. The linkage was conducted using the Boolean operator (AND, OR). The choice of keywords was intended to be broad to maximise the number of relevant studies considered. The following search strategy was applied to Ovid and PubMed:

Furthermore, the following search strategy was applied to Scopus and Web of Sciences to supplement records identified through Ovid and PubMed:

(non steroidal anti inflammatory agent OR non steroidal anti inflammatory agents OR non steroidal anti inflammatory drug OR non steroidal anti inflammatory drugs OR cyclooxygenase inhibition OR COX inhibition OR ibuprofen OR celecoxib) AND (osseointegration OR osteoblast OR osteoblasts OR titanium implant OR titanium implants OR dental implant OR dental implants)

The titles and abstracts of each research article in the initial search were independently screened by the primary (JDL) and the second reviewer (TJ). A third independent reviewer (MN) reviewed the articles that were included by one but excluded by the other reviewer. The full-text manuscripts of the articles were catalogued in accordance to the “Eligibility criteria” section as mentioned above and were assessed for the following criteria:

The full-text manuscripts of included studies were catalogued into in vitro, clinical, and in vivo studies. The data from the included studies were independently extracted by the primary (JDL) and the second reviewer (TJ) according to the “Data items” section as listed below. Disagreements or uncertainties were discussed with the third reviewer (MN) until an agreement was reached.

The data collected from the included studies were arranged in the following fields:

The quality and risk of bias assessments were performed independently by two reviewers (JDL and TJ) during the data extraction process. Any disagreements or uncertainties were discussed with the third reviewer (MN) until an agreement was reached. The quality and bias assessment for all studies addressed various bias domains. A Modified CONSORT checklist of items for reporting in vitro studies of dental materials outlined by Faggion (Tables 1 and 2) was used to assess quality and risk of bias of included in vitro studies [16]. The Cochrane Collaboration’s Tool (Table 3) was used to assess quality and risk of bias of included human clinical studies [17]. A quality assessment of the methodology of the animal studies (Table 4) has been performed according to items (Table 5) of the ARRIVE guidelines [18].

The relevant data collected for qualitative synthesis are summarised in three critical analysis tables: in vitro studies (Table 6), clinical studies (Table 7), and in vivo studies (Table 8).

No meta-analyses could be performed due to the heterogeneity between the studies—different samples, experimental groups (drugs and concentrations), study models, and outcome measures.

The electronic search in the databases of Web of Science, PubMed, Ovid, and Scopus resulted in the identification of 875 potential titles and abstracts. After removal of the duplicates, independent screening of the abstracts resulted in the selection of 79 studies for assessment of eligibility. A total of 13 studies were eligible and are included in the systematic review (Fig. 1).

The eligibility and study selection criteria as mentioned above were applied to the 79 full-text articles. A total of 66 studies were excluded after a full-text assessment for the following reasons:

The included studies were catalogued into three groups characterised by the type of study: in vitro studies (Table 6), clinical studies (Table 7), and in vivo studies (Table 8). The cataloguing provided a clearer understanding of the effects of NSAIDs in osseointegration in various study models, ultimately contributing to the sensitivity of the systematic review.

The quality and risk of bias assessments of included studies are summarised in Tables 1, 2, 3, 4, and 5. The quality assessment revealed a high risk of bias (for one or more domain) for most of the included studies. The included in vitro studies had high risk of bias according to the Modified CONSORT checklist [16, 19, 20]. One clinical study was classified as unclear risk of bias (for one or more domain) according to the Cochrane Collaboration’s Tool [17, 21]. An in vivo animal study had an unclear risk of bias according to the ARRIVE guidelines [18, 22].

The effects of NSAIDs on the osteogenic activity of osteoblasts have been extensively studied at the molecular pharmacological level [23]. However, only two studies have been identified that investigated the effect of NSAIDs on osteoblasts attached to titanium surfaces (Table 6). In the study conducted by Boyan et al., their results demonstrated that a non-selective COX inhibitor (indomethacin, 0.1 μM), a selective COX-1 inhibitor (resveratrol, 1 and 10 μM), and a selective COX-2 inhibitor (NS-398, 1 and 10 μM) did not have a significant effect on the number of cells derived from human osteosarcomas [20]. Furthermore, Boyan et al. demonstrated that the NSAIDs reduced prostaglandin E₂ (PGE2) production of cells attached to a rough titanium surface. Their results indicated that both COX-1 and COX-2 are involved in the production of osteocalcin, PGE2, and TGF-β1 by osteoblasts [20]. They also demonstrated that osteoblasts produced increased levels of PGE2 on rough titanium surfaces and that this was correlated with increased alkaline phosphatase activity and osteocalcin production [20]. This suggests that PGE2 may have a role in osteoblast proliferation and differentiation on titanium surfaces, and that this favourable effect of PGE2 was inhibited when a NSAID was present [20]. Arpommaeklong et al. found that a non-selective COXinhibitor (indomethacin, 0.1 μM) and a selective COX-2 inhibitor (celecoxib, 1.5, 3.0, and 9.0 μM) inhibited the growth of cell cultures derived from rat calvarias, where the effect was dose-dependent in the cultures treated with celecoxib [19]. Furthermore, Arpommaeklong et al. demonstrated that PGE2 levels were significantly lower in the groups that were treated with a NSAID, and have postulated that PGE2, consistent with Boyan et al., may have a role in osteoblast growth and differentiation [19, 20].

The clinical evidence demonstrating the effects of NSAIDs on the osseointegration of titanium dental implants is limited with only two clinical trials and one retrospective study identified in the database searches (Table 7). In the clinical trial conducted by Alissa et al., the effect of a 7-day post-operative course of ibuprofen (600 mg, taken four times daily) on the marginal bone level around dental implants was investigated [21]. They found that there were no statistically significant differences between the treated and placebo groups in the mean marginal bone level around dental implants at 3 and 6 months after implant placement [21]. In a similar clinical trial conducted by Sakka et al., the effect of a 7-day course of ibuprofen (600 mg, taken four times daily) on the marginal bone level around dental implants was also investigated [24]. They found that there were no significant differences between the ibuprofen and non-ibuprofen groups, consistent with the findings of Alissa et al., when comparing the changes in marginal bone level around dental implants [24]. However, a retrospective study conducted by Winnett et al. postulated that the adverse biological events following dental implant placement were associated with perioperative use of NSAIDs [25]. Winnett et al. reported a total loss of 197 dental implants due to failed osseointegration from patients with failing implant/s (468 implants in 104 patients) treated in a postgraduate dental clinic (between 1979 and 2012). The patients (n = 60) that used NSAIDs peri-operatively experienced a total of 119 failed implants, whilst the non-NSAID cohort (n = 44) experienced a total of 78 failed implants. Winnett et al. identified that ibuprofen (600 mg, taken four times daily) was the most commonly prescribed; however, other prescribed NSAIDs included Ketorolac, Vioxx, Celebrex, Diflunisal, Meloxicam, and Naproxen [25]. Despite the clinical trials conducted by Alissa et al. and Sakka et al., both of whom have demonstrated that a 7-day post-operative course of ibuprofen (600 mg, taken four times daily) did not significantly affect bone levels around dental implants at 3 and 6 months after placement, the data gathered by Winnett et al. indicates that NSAIDs may have detrimental effect on osseointegration [25].

The influence of NSAIDS on bone healing in animal models has been shown to be related to the duration of treatment and drug selectivity [5]. A total of seven studies were identified that investigated the effect of NSAIDs on the osseointegration of titanium implants in animals: mice, rabbits, and rats (Table 8).

The duration of treatment is a factor to consider when using NSAIDs, and a study conducted by Goodman et al. investigated the effect of a selective COX-2 inhibitor (rofecoxib, 12.5 mg/kg/day) administered for 6 weeks on bone growth in a bone harvest chamber during three different time periods: initial 2 of the 6 weeks, final 2 of the 6 weeks, and continuously for 6 weeks [26]. The bone harvest chamber was a titanium device that was implanted into the tibia of rabbits and had an inner removable core with canals that allowed for bone ingrowth into the inner chamber. Their results revealed that rofecoxib given continuously for 6 weeks significantly reduced bone ingrowth and osteoblasts per area compared with the control (no treatment), whilst rofecoxib given for 2 weeks did not appear to interfere with bone ingrowth and number of osteoblasts [26]. Furthermore, the studies conducted by Ribeiro et al. investigated the effect of long-term administration (60 days) of a selective COX-2 inhibitor (meloxicam, 3 mg/kg/day) on the bone growth on a titanium implant [27, 28]. Their results also indicated that long-term use of a selective COX-2 NSAID significantly reduces bone-to-implant contact, bone area, and bone density, ultimately leading to failed osseointegration [27]. The data gathered in both studies suggest that duration of treatment is an important factor in the use of selective COX-2 NSAIDs, as short periods of rofecoxib and meloxicam did not adversely affect osseointegration [26, 27].

The COX selectivity of NSAIDs and their interference with prostaglandin synthesis have been shown to inhibit bone healing [23]. Goodman et al. performed a follow-up study where the titanium bone harvest chamber was again implanted into the tibia of rabbits and the rabbits were treated with water (control), a non-selective COX inhibitor (naproxen, 110 mg/kg/day), or a selective COX-2 inhibitor (rofecoxib, 12.5 mg/kg/day) for a 4-week period [29]. Their results again demonstrated that COX-2 inhibition significantly decreased bone ingrowth, where rofecoxib also decreased the number of osteoblasts per area [29]. Furthermore, the conclusions by Goodman et al. were supported by the study by Chikazu et al. where titanium implants were inserted in wild-type and COX-2-knockout mice [30]. Their results revealed that the expression of COX-2 was induced in bone surrounding the implants in wild-type mice, but not in COX-2-knockout mice and that the bone-to-implant contact was minimal in newly formed bone in COX-2-knockout mice [30]. The data collected by Goodman et al. and Chikazu et al. postulated that COX-2 may have an important role in osseointegration [26, 30]. However, a study conducted by Pablos et al., that investigated the effect of a non-selective COX inhibitor (diclofenac, 1.07 mg/kg/day) and a selective COX-2 inhibitor (meloxicam, 0.2 mg/kg/day) administered for 5 days on peri-implant healing in rats, revealed that diclofenac delayed peri-implant bone healing and negatively affected the bone-to-implant contact, whereas meloxicam had no negative effect on peri-implant healing [31]. The results of Pablos et al. were inconsistent with the results of the study conducted by Cai et al. that also investigated the effect of diclofenac (2 mg/kg/day) and a selective COX-2 inhibitor (parecoxib, 1.5 mg/kg/day) administered for 7 days on the osseointegration of titanium implants in rabbit calvarias. Their results revealed no statistically significant differences between the experimental groups and the control [22]. Furthermore, a recent study conducted by Salduz et al. that investigated the effect of a non-selective COX inhibitor (diclofenac, 5 mg/kg/day) and a selective COX-2 inhibitor (celecoxib, 3 mg/kg/day) administered for 8 weeks on bone growth and osseointegration on two different alternative titanium surfaces revealed no statistically significant differences in the biomechanical and histological results between the experimental groups and the control, suggesting that long-term use of NSAIDs does not affect osseointegration [32]. The data collected in animal studies regarding duration of treatment and drug selectivity is inconsistent, and there is a lack of consensus on the influence of NSAIDs on osseointegration in animal models.

The majority of the included studies revealed a high risk of bias, and conclusions from studies that have a high risk of bias are sufficient to affect interpretation of data [16‐18]. Publication and selection bias is apparent in several included studies, as the negative effects of NSAIDs on osseointegration can be expected in the studies that administered NSAID at a high concentration and/or for a prolonged period of time. The conclusions of this systematic review were largely based on animal studies, as there are very few published in vitro and clinical studies relating to the effect of NSAIDs on osseointegration. The effects of NSAIDs on osseointegration in animals cannot be translated to humans due to the vastly different pharmacokinetics.