Transalveolar sinus floor lift without bone grafting in atrophic maxilla: A meta-analysis

We performed a meta-analysis aimed to assess the clinical results after transalveolar sinus floor lift without bone grafting in the atrophic maxilla. A systematic electronic literature search was conducted in PubMed, Embase and The Cochrane Library, followed by a manual search. Two reviewers independently extracted study data and conducted quality assessments. Ten non-controlled studies including 1484 implants and eight controlled studies (5 RCTs and 3 prospective studies) including 817 implants (451 implants in the non-graft group) were enrolled in this study. The survival rate of implants via the graft-free method was 98% (95%CI 96% to 100%). There was no significant difference in the survival rate between the non-graft group and the graft group (RR: 1.02; p = 0.18). No statistically significant difference in marginal bone loss was detected between the groups at 12 months (0.57, p = 0.07) or 36 months (0.05, p = 0.61). The endo-sinus bone gain in the non-graft group was significantly lower than in the graft group at 12 months (−1.10, p = 0.0001) and 36 months (−0.74, p = 0.02). Hence, the available evidence suggests that predictable results could be acquired through transalveolar sinus floor lift without bone grafting, while there may be a trend toward more endo-sinus bone gain with bone grafts.


Results
Study identification and selection. The flowchart for study identification and selection is shown in Fig. 1. There were 2601 studies identified after the initial electronic search. Seven hundred and nine duplicate studies were excluded, and 1892 records remained. One thousand seven hundred and seventy-five articles were excluded after reviewing the titles and abstracts. The remaining 117 full-text articles were further evaluated for eligibility. Ninety-seven of them were also excluded due to not fulfilling the inclusion criteria or meeting the exclusion criteria, while two additional articles were identified by reviewing the references of the full-text studies. Ultimately, eighteen studies [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27] , including 10 non-controlled studies and 8 controlled studies, were included in this meta-analysis.
Characteristics of included studies. Among the fifteen included studies, there were 10 non-controlled studies assessing graft-free methods involving 1484 implants. Another 8 controlled studies comprised a total of 817 implants, including 451 implants in the non-graft group and 366 implants in the control group. The selected studies were published between 2008 and 2017 and were conducted in China, Canada, Spain, Sweden, Germany, Italy, Turkey and Switzerland. The main characteristics of the eighteen included studies are presented in Tables 1 and 2A. The outcome data for each included controlled trial are presented in Table 2B.
Quality assessment of included studies. We assessed the risk of bias of the 5 RCTs and 13 non-RCTs using a risk-of-bias assessment tool (Fig. 2).
For the 5 RCTs, the blinding of the participants and personnel was unclear in 3 trials. Two studies 22,23 were identified as being "Low Risk Bias". The risk of bias for the other 3 studies 19,20,24 was unclear.
For the other 13 non-RCTs, all ten non-controlled studies acquired MINORS scores between 9 and 12. The remaining three uncontrolled studies acquired MINORS scores of 20. The methodological quality assessment is displayed in Fig. 2b.

Analysis of outcomes. Primary outcomes.
1. Survival rate of implants in non-controlled studies (without grafting) Survival rates for implants were available in all 10 included studies with the longest follow-up at 120 months. The meta-analysis was conducted using R software (version 3.1.3). As significant heterogeneity among the studies was detected (I 2 = 78.8%, p < 0.0001), a random-effects model was selected for a more conservative effect. The pooled analysis showed that the survival rate of the graft-free method was 98% (95% confidence interval 96-100%) (Fig. 3a). 2. Survival rate of implants in controlled studies (non-graft group vs. graft group) Eight controlled studies, including 5 RCTs and 3 prospective studies, reported the survival rates in the non-graft group and graft group. As three RCTs 19,20,24 included the same samples, we selected the longest follow-up data 24 . In the end, six studies were included in the pooled analysis. The pooled analysis was conducted using Review Manager software (version 5.2, Cochrane Collaboration, Oxford, UK). A risk ratio of 1.02 (95% confidence interval 0.99-1.05) was found in the meta-analysis. There was no statistically significant difference between the two groups (P = 0.18), and no heterogeneity among the studies was detected (I 2 = 0%, p = 0.55) (Fig. 3b).

Survival rates of implants in different RBH for transalveolar sinus floor lift without grafting
Six of the ten non-controlled studies reported the RBH when the survival rates of the implants were available. Four studies included patients with RBH ≤ 4 mm and RBH > 4 mm. The survival rate at RBH ≤ 4 mm was 95.35%, while it was 96.34% for RBH > 4 mm. Two studies reported survival rates for patients with RBH ≤ 5 mm (95.18%) and RBH > 5 mm (95.12%) ( Table 3).
Secondary outcomes.

Study
Year Study type Country Publication bias. No evidence of publication bias was detected, as all the outcomes had funnel plots with no significant asymmetry (Fig. 5).

Discussion
Si MS et al. 22 found a survival rate that was slightly higher in the graft group than in the non-graft group. Other researchers 18,21,24 found that the survival rates in the graft group were slightly lower than in the control group. However, none of these studies showed a statistically significant difference in the survival rate between the two groups. The marginal bone loss at 12 months and the endo-sinus bone gain at 36 months showed statistically significant differences in different studies 19,20,22 . As most of these studies had small sample sizes, our meta-analysis might help to achieve more reliable results. To minimize the potential for bias, the reviewers extensively searched the published literature through a search engine (PubMed) and electronic databases (Embase and The Cochrane Library) and manually retrieved the references of the included studies. All the included studies were strictly enrolled based on the inclusion and exclusion criteria. Data extraction and quality assessment of the studies were performed by two reviewers, respectively. Ultimately, eighteen studies including five RCTs were included. The sample size for implants reached 1484 implants in the non-controlled studies and 817 implants in controlled studies. The results of our meta-analysis indicate that a graft-free sinus floor lift had a positive effect with a survival rate of 98% (95%CI 96-100%). This study showed good survival rates for implants in both cases with and without bone grafting after sinus floor lift. A slight survival benefit was detected in the non-graft group, but no statistically significant difference was observed (RR = 1.02, p = 0.18). The overall marginal bone loss at 12 months and 36 months was slightly higher in the graft group but the difference was again not statistically significant.
The studies included in our meta-analysis reported endo-sinus bone gain in the non-graft group, which was in accordance with prior studies suggesting that new bone could form under the lifted sinus membrane as a result of the physical space and blood clot formation 28,29 . The results of our meta-analysis suggest that the overall amounts of endo-sinus bone gain at 12 months and 36 months were lower in the non-graft group, and the difference was statistically significant. The differences were 1.10 mm at 12 months and 0.74 mm at 36 months.  There may be a trend for more endo-sinus bone gain when using grafting materials. Future studies could focus on the issue of whether this difference is the result of more bone resorption or simply less membrane elevation. Moreover, the graft shrinkage rate is also important. As only one study reported on post-OP X-rays in the graft group, a meta-analysis of the graft shrinkage rate could not be conducted.
Differences in the RBH might influence the survival rate of implants. The survival rate of implants for RBH ≤ 4 mm was only slightly lower than for RBH > 4 mm. However, different results were detected for RBH ≤ 5 mm compared with RBH > 5 mm, which could be attributable to the limited sample size.
Although no study reported a cost/effectiveness ratio, it is clear that the cost of a sinus floor lift without bone grafting must be lower than with grafting.
There are some limitations to this meta-analysis. First, certain issues that might potentially influence the results of the included studies should be clarified. The studies involve different study designs, initial RBHs, implant lengths, grafting materials, measurement instruments and implant types. In addition, the anatomical defect and need for grafting associated with surgical access and surgical technique may also influence the results of the included studies. Most studies measured outcome data by periapical radiograph or panoramic radiograph, while only three studie 14,23,27 used cone beam computed tomography (CBCT). Second, there was a limited number of studies of different quality levels included. There were only 5 RCTs included, of which three 19,20,24 used the same samples. Two RCTs had a low risk of bias, and three RCTs had a moderate risk of bias. The other studies were all non-RCTs. Although it is hard to quantify the influence of these risks of bias on the study results, such methodological shortcomings should be considered when interpreting the results of our study. Third, only articles published in English were included, which means that potentially relevant articles published in other languages might not have been identified. Excluding these studies might contribute as a potential source of bias in this study.
In conclusion, both graft-free sinus floor lifts and procedures with grafts have positive effects; the results of our meta-analysis indicate that there are no differences in survival rate and marginal bone loss between non-graft and graft groups; the endo-sinus bone gain in the graft group was slightly higher than in the non-graft group. However, because of the limitations of our study, as mentioned above, future well-designed RCTs with long-term follow-up are also required to substantiate our findings.

Material and Methods
This study followed the guidelines of the PRISMA statement (http://www.prisma-statement.org/).

Literature search. A systematic electronic literature search was conducted in PubMed, Embase and The
Cochrane Library (all from inception to October 2017). The medical subject headings (MeSH) "sinus floor augmentation" and "bone transplantation" and the text words "sinus lift", "sinus augmentation", "sinus floor elevation", "sinus elevation", "bone graft*", "graft*", "bone augmentation", "graft-free", and "non-graft*" were used in combination with other strategies to identify potential studies. The publication language was restricted to English. To be as inclusive as possible, no limitations were set for the Design, Region, or Publication type. Moreover, a manual search of all the relevant references in the included studies was performed to discover other potentially eligible trials. This process was conducted iteratively until no additional trails could be identified. To minimize the potential for reviewer bias, two reviewers independently conducted electronic literature searches and performed study selection. The level of agreement between the reviewers was determined by the Cohen k test, assuming k = 0.61 as an acceptable agreement score 30 . Any disagreement regarding inclusion or exclusion of a retrieved study was resolved by discussion or consulting another reviewer.   Quality assessment. The two reviewers independently assessed the risk of bias for each included study, and all disagreements were resolved by consensus or consulting a third reviewer. RCTs were evaluated using the Cochrane Collaboration's Risk of Bias tool (www.cochrane-handbook.org); when there was more than one item with "Unclear risk of bias" or "High risk of bias", the quality of the study was considered "Unclear risk of bias" or "High risk of bias". Non-RCTs (prospective studies and retrospective studies) were assessed using the Methodological Index for Non-Randomized Studies (MINORS) 31 . The methodological quality was scored from 0 to 16 for studies without a control group and from 0 to 24 for studies with a control group.

Statistical analysis
Meta-analysis of the survival rate in non-controlled studies was performed using R software (version 3.1.3), while the other meta-analyses were conducted using Review Manager software (version 5.2, Cochrane Collaboration, Oxford, UK). We used the mean difference (MD) and risk ratio (RR) to compare continuous and dichotomous variables, respectively. All measures were reported with 95% confidence intervals (CIs). Statistical heterogeneity was detected using the I 2 statistic and the chi-squared test. First, a fixed-effects model was used. When significant heterogeneity (I 2 > 50%) was detected, we changed to a random-effects model 32 . Subgroup analyses and sensitivity analyses were carried out when necessary to determine whether there was a difference in the results. Furthermore, the risk of publication bias was investigated for the included outcomes by analysing funnel plots through visual inspection 33 .