Clinical and Radiographic Variables Around Implant With Simultaneous Graft Among Type 2 Diabetic Patients Treated With Different Hypoglycemic Medications: A Retrospective Study.

Background: The clinical and radiographic variables around dental implants in type 2 diabetes mellitus patients with different hypoglycemic agents still remained unclear. Methods: This retrospective cohort study collected the dental records and digital periapical radiographs of type 2 diabetes mellitus patients and implants. These patients were grouped according to their medication: insulin, metformin, and glucagon-like peptide-1 drugs. The radiographic marginal bone loss around implants and clinical parameters, including peri-implant bleeding on probing and probing depth, were compared among groups using the Kruskal-Wallis test. Results: A total of 150 patients with 308 implants (101 in insulin group, 121 in metformin group and 86 in glucagon-like peptide-1 drugs group) were assessed. The peri-implants marginal bone loss in insulin group (P<0.05) and metformin group (P<0.01) were signicantly higher than glucagon-like peptide-1 drug group. The radiographic bone loss in metformin was higher than insulin group (P<0.05). While there was no statistical difference of clinical peri-implant parameters among groups(P>0.05). Conclusions: The radiographic variables were not exactly the same among type 2 diabetes mellitus patients with different hypoglycemic agents. glucagon-like peptide-1 drugs might be more benecial to bone tissue around implants. More studies are needed to verify the direct effect of these drugs on peri-implant bone. and GLP-1 (group 3). The state immediately after surgery was recorded as baseline. The MBL from immediately after implant placement to before prosthetic installation and to the rst-year follow-up after functional loading were measured to suggest the change of bone tissue around implants. The clinical inammatory parameters at the rst-year follow-up after functional loading, including bleeding on probing (BOP) and probing depth (PD), were collected. To ensure the principle of blinding, the groups were sealed and another independent examiner assessed the data among groups.


Background
The number of individuals with type 2 diabetes mellitus (T2DM) constitutes a large population, and a recent report in 2019 showed that, the prevalence of T2DM worldwide had reached 8% in 2018 1 . Due to the relationship between T2DM and periodontitis 2 , there is a large proportion of patients with missing teeth in the T2DM population. Implant-supported denture restoration has been an excellent treatment for the loss of teeth and T2DM individuals have a large requirement for dental implant treatment. However, T2DM patients often face the challenge of atrophic alveolar bone width at implant sites 3 . The minor and moderate atrophic ridge often need horizontal bone augmentation by guided bone regeneration (GBR)with the combination of bone graft materials and barrier membranes 4 . The uncontrolled hyperglycemia would impact the therapeutic effect of dental implants in T2DM 5 . And some studies have shown that early osseointegration around implants is compromised by T2DM even if glucose is strictly controlled 6 . It is crucial for clinicians to promote the e cacy of implant treatment in T2DM patients.
Dental implant treatment success is related to the osseointegration and peri-implant bone around implants [7][8] . Previous research has shown that diabetes impacts bone remodeling around implants, especially at the early stage of osseointegration 9 . The changes of bone levels around implants is a biomechanical process that could be in uenced by several factors in patients and implants 10 . The marginal bone loss (MBL) in diabetes patients is higher than that in nondiabetic individuals, despite being glycemic controlled 11 . The higher MBL in T2DM patients could re ect the impaired bone condition around implants. Therefore, MBL was considered an important radiographic parameter for bone tissue around implants in the present study.
Recently, abundant evidence has revealed the role of hypoglycemic agents in bone metabolism 12 . There is growing concern about their direct bone target effect. For example, insulin and metformin could be bene cial to bone tissue 13 . This bene cial effect might improve the outcome of dental implants. It was reported that local application of insulin could promote osseointegration in T2DM rats 14 . Glucagon-like peptide-1 (GLP-1) drugs also have the potential to promote osseointegration around implants 15 . Currently, there is no clinical evidence of hypoglycemic drugs interfering with bone tissue around implants. Therefore, the present study aimed to obtain more detailed information about the clinical and radiographic variables around implants in T2DM patients using different hypoglycemic agents.

Ethical protocol
The retrospective cohort study was approved by the Ethics Committee of School of Stomatology, the Fourth Military Medical University (Ethics Approval Number: IRB-REV-2020045) and in compliance with the Helsinki Declaration. All patients were given complete information about the treatment and signed informed consent forms before surgery. The implant treatment was performed by the same doctor and the trauma was minimized. This study is reported in agreement with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement.

Study designand setting
This retrospective cohort study was designed to compare the clinical and radiographic variables around implants with simultaneous GBR in T2DM patients using different hypoglycemic agents. The examiners collected dental records of T2DM patients who received implant surgery from January 2015 to November 2019. After information collection, one examiner divided them into three groups based on their major hypoglycemic agent species: insulin (group 1), metformin (group 2), and GLP-1 (group 3). The state immediately after surgery was recorded as baseline. The MBL from immediately after implant placement to before prosthetic installation and to the rst-year follow-up after functional loading were measured to suggest the change of bone tissue around implants. The clinical in ammatory parameters at the rstyear follow-up after functional loading, including bleeding on probing (BOP) and probing depth (PD), were collected. To ensure the principle of blinding, the groups were sealed and another independent examiner assessed the data among groups.

Eligibility criteria
The inclusion criteria were as follows: (a) medically diagnosed T2DM; (b) good blood glucose control before surgery (HbA1c%≤8% 16 ); (c) treatment with one of the hypoglycemic agents: metformin, insulin or GLP-1; (d) implant site with Seibert class I ridge de ciencies resulting in the implant thread exposure which treated only with GBR (graft mass≤0.25g); (e)good oral hygiene care with regular semidiurnal brushing and semiannual professional cleaning.
The exclusion criteria were as follows: (a) poor glycemic control (HbA1c%>8%) or severe diabetic complications; (b) osteoporosis; (c) long-term use of bisphosphonates or steroids; (d) irregular use of hypoglycemic drugs; (e) excessive tobacco use and alcohol consumption; (f) progressive periodontitis.

Information collection
The dental records of participants were reviewed by examiners and the baseline information about patients and implants were obtained. Patient information included age, sex, HbA1c, medication, bone augmentation and oral hygiene care. The implant information included the size, location, arch, position of xture and major surface treatment of implants. The surgical complications after implant placement were recorded, including wound bleeding, swelling and membrane exposure. And the infection around implant was obtained.

Radiographic analysis
A standardized digital dental periapical radiographic evaluation performed immediately after implant placement was recorded as the baseline. The radiographic evaluation was carried out before prosthetic installation and at 1-year follow-up after function loading ( Figure 1).
All radiographs were viewed on a computer screen using Digimizer v5.4.5 (MedCalc Inc., Mariakerke, Belgium) software. The examiner measured the crestal bone levels as the vertical distance from the tip of the implant body to the coronal edge of the rst bone-to-implant contact ( Figure 2). The image error of magni cation was calibrated by the reference value of the actual implant length.

Clinical parameters at the rst year after functional loading
The clinical parameters recordings at the rst year after functional loading were collected and analyzed, including BOP and PD. These clinical parameters were measured at six sites around implants and reported as the mean value.

Statistical analysis
Statistical analysis was performed using specialized software (SPSS v25.0, IBM, Chicago, Illinois). The baseline information was analyzed by the chi-square test and analysis of variance. Dependent variables were evaluated for a normal distribution by the Shapiro-Wilk test. The data are expressed as the means and SDs. Peri-implant MBL and clinical parameters was compared using the Kruskal-Wallis test. The Bonferroni post hoc test was applied for multiple comparisons. Multiple regression analysis was performed to control the baseline variables as covariates. The direction and strength of association between outcome variable and covariates were assessed by regression coe cients, and the precision of which was measured by 95% con dence intervals. Statistical signi cance threshold was set at P< 0.05.

Results
The general condition of population A total of 150 patients with 308 implants with GBR were eligible for this study. The system of implant included Straumann, Basel, Switzerland and Nobel Biocare, Gothenburg, Sweden. The condition of implant was analyzed by position of xture and the major surface treatment to avoid the con icts of interest. Sixty-one implants in groups with GBR experiences surgical complication and no implant suffered infection. And 54 patients with 101 implants were followed up for a year after the prosthesis installation, and other patients lost to follow-up. No further complications were reported during the followup period.

The Information Analysis Of Patients And Implants With Gbr
The basic information about patients and implants with GBR is shown in Table 1. There were 54 patients and 101 implants in insulin group, 54 patients and 121 implants in metformin group, 42 patients and 86 implants in GLP-1 group. These groups were comparable with respect to age, HbA1c, sex, implant location and the major surface treatment of implants (P > 0.05). However, the position of xture was signi cantly different among the groups (P < 0.01). A total of 71 patients with 129 implants completed one-year follow-up after functional loading (30 patients and 53 implants in insulin group, 24 patients and 44 implants in metformin group, 17 patients and 32 implants in GLP-1 group). The basic confounding factors of patients and implants among groups were comparable (P > 0.05). Notes; Group 1: insulin. Group 2: metformin. Group 3: GLP − 1. GBR, guided bone regeneration. a: the position of xture among groups is signi cantly different (P < 0.01). **p < 0.01.
The comparison of radiographic and clinical parameters of implants with GBR At 12 months of follow-up, the mean MBL around implants with GBR is shown in Table 2. The MBL was not the same among these medication groups. The mesial and distal MBL parameters in group 1 (mesial: 0.43 ± 0.14 mm, distal: 0.42 ± 0.13 mm) (P < 0.05) and group 2 (mesial: 0.45 ± 0.14 mm, distal: 0.47 ± 0.13 mm) (P < 0.01) were higher than those in group 3 (mesial: 0.38 ± 0.12 mm, distal: 0.36 ± 0.12 mm). The distal MBL was signi cantly higher in group 2 (0.47 ± 0.13 mm) than in group 1 (0.42 ± 0.13 mm) (P < 0.05). Similarly, the distal MBL in the percentage of implant length was signi cantly higher in group 1 (P < 0.01) and group 2 (P < 0.01) than in group 3. While at 24 months of follow-up, the mesial and distal MBL parameters in group 1 (mesial: 0.63 ± 0.15 mm, distal: 0.61 ± 0.17 mm) (P < 0.05) and group 3 (mesial: 0.58 ± 0.13 mm, distal: 0.61 ± 0.19 mm) (P < 0.05) were smaller than those in group 2 (mesial: 0.72 ± 0.14 mm, distal: 0.68 ± 0.13 mm).  Regression analysis to control the position of xture at the 12months of follow-up The position of xture among groups should be controlled as confounders by multiple regression analysis ( Table 4). The mesial and distal MBL at 12 months of follow-up still showed statistically signi cant differences among the groups even after controlling for the position of xture (P < 0.05). The table IV showed that the mesial MBL at 12 months of follow-up in group 1 (P < 0.05) and group 2 (P < 0.01) were higher than group 3 and the distal MBL in group 1 (P < 0.01) and group 3 (P < 0.01) were smaller than group 2 after controlling the position of xture consistent among groups. This study showed that MBL among groups was not exactly the same. The MBL parameters in insulin group and metformin group were higher than those in GLP-1 group at 12 months of follow-up. The MBL in metformin group was higher than insulin group and GLP-1 drugs group at 24 months of follow-up. Similarly, the MBL in the percentage of original implant length in insulin group and metformin group was higher than that in GLP-1 drugs group. During these periods, the bone and bone substitution material had experienced the bone healing and remodeling process and this process could be in uenced by the bone condition around implants. The difference of MBL among groups showed the possibility of these drugs interfering with bone tissue around the implant in T2DM patients. After controlling for the baseline information variable, the difference in MBL among groups still existed. This means that these confounding variables, including position of xture and major surface treatment, did not in uence the comparative results of MBL among groups. The change of bone levels around implants could show the bone healing and remodeling process and the higher MBL might suffer higher risk of failure. From these results, the medication type might be related to the bone healing and remodeling around implant at the early stage and the potential among medication might be different. However, the present result showed there was no statistically signi cant difference in clinical parameters, including BOP (+) and PD, among groups. Although the diabetes could be related to the periodontal in ammation, the peri-implant in ammation parameters seem to have nothing to do with the medication type of the T2DM patient.
The results showed that GLP-1 drugs might have a more positive effect on the peri-implant bone than other agents. This is consistent with previous studies. Liang 17  Metformin and insulin are both classic hypoglycemic medications and there are many studies about their positive in uence on bone tissue. The results showed that the distal MBL in the metformin group was higher than that in the insulin group (P < 0.05). The most previous comparison of the bone metabolism effect between these two drugs focused on the systematic body. Raj 24 pointed out that compared with metformin, insulin could signi cantly protect bone through osteocalcin and other pathways. However, some fracture risk studies presented different views. Losada 25 and Hidayat 26 determined that metformin could reduce the fracture risk and protect bone metabolism compared with the effects of insulin. The underlying mechanism of this contradiction is unclear. The explanation may be that diabetes is generally more severe in insulin users than oral medicine users. Patients using insulin could suffer more fall fractures caused by illness complications and hypoglycemic reactions, while metformin users had a lower incidence of fractures due to fewer complications 27 . Therefore, the fracture risk cannot accurately re ect the in uence of drugs on bone metabolism. This study selected patients with good blood glucose control and no severe complications. The interference of different conditions was removed as far as possible to increase the reliability of the results.
There are also many studies on the bene cial bone target mechanism of metformin and insulin. A large number of studies have proven the positive effect of insulin on bone tissue. Insulin has been shown to promote bone formation by upregulating the serum osteogenesis factor 28 . Some studies have shown that a local injection of insulin can promote early fracture healing in diabetic animals 29 . Moreover, local or systematic use of insulin could improve implant osseointegration [30][31] . Another study has shown that insulin could promote angiogenesis 32 , which is also a conducive factor to early osseointegration.
Metformin is a rst-line drug for the treatment of diabetes. Its positive mechanism on bone has been con rmed by a large number of preclinical studies. It is reported that metformin could promote osteogenesis and inhibit bone resorption 33 . However, metformin would reduce the level of bone turnover factors and hinder the bone remodeling 34 . This effect might compromise the self-repair of minor bone tissue damage. In this study, the MBL in the insulin group was smaller than that in the metformin group, which might be related to their effect on bone metabolism.
At present, the major controversy about the bone target effect of some hypoglycemic drugs entails their indirect effect on bone by controlling blood glucose 35 . Firstly, it is noteworthy that all T2DM patients included in this study had good controlled blood glucose, with HbA1c less than 8%. And the results showed that MBL was different among groups, even though the patients in each group reported good controlled blood glucose. This meant that these drugs might have different in uences on peri-implant bone. The process of bone remodeling around implants requires an environment conducive to bone formation. Therefore, T2DM patients undergoing implant surgery should choose hypoglycemic drugs more conductive to bone formation, such as insulin and GLP-1RA. However, this adjustment in the medication plan should not violate the clinical principle of diabetes treatment. Hence, the results of this study are more useful for patients who have exibility regarding agent use. One limitation of this study is that blood glucose could not be monitored in real time. Therefore, further laboratory-based investigations should be executed to guarantee blood glucose control more strictly and eliminate bias in this type of research in the future.
Another contribution of the present study is that it provides feedback on clinical systemic bone research. Previous research on the systematic bone target effect of hypoglycemic drugs focused more on the changes in fracture risk, bone mineral density and bone-related biochemical indexes 36 . However, these indexes could not accurately re ect the in uence of drugs on bone. The bone mineral density in patients with T2DM might be higher than that in normal individuals, and the fracture risk in T2DM patients could be high 37 . The fracture risk reason might be related to damaged bone quality, higher incidence of falls and severe complications of diabetes. While the MBL around the implant is an important criterion for the e cacy of implant treatment and could be affected by bone metabolism 38 . Therefore, MBL could serve as an indicator of the bone target effect and could be used to supplement relevant research results about systematic bone.
Considering the cost and radiation risk, conventional standard periapical radiography is still the most preferable clinical method to assess MBL around implants. This study focused on the MBL around implants with GBR in different hypoglycemic agent groups. However, it is beyond the scope of this study to examine the horizontal change of the alveolar ridge at the implant site as this study was based on twodimensional images. Follow-up studies on the horizontal change in the ridge would be executed by cone beam computed tomography.

Conclusion
Though some studies have shown the different effects of medication on bone metabolism in T2DM patients, there is still a lack of clinical investigation for peri-implant parameters in T2DM patients with different hypoglycemic agents. This study could offer new reference regarding medication for T2DM patients undergoing implant surgery. The ndings of this study suggest different MBL around implants in T2DM patients using different hypoglycemic agents. GLP-1RA seems be bene cial to bone tissue around the implants compared with insulin and metformin. However, the present study could not provide a high level of evidence for the in uence of these drugs on bone condition around implants. Because the skeletal effects of hypoglycemic drugs are very complex, and much clinical evidence is still needed through future research.