A Scoping Review of the Use of Pioglitazone in the Treatment of Temporo-Mandibular Joint Arthritis

Thiazolidinediones (TZDs) are a group of diabetes medications currently being investigated for anti-arthritis effectiveness, one of which is pioglitazone. The purpose of this scoping review is to evaluate the potential use of pioglitazone in the treatment of temporomandibular joint (TMJ) arthritis. The criteria of eligibility were studies with the diagnosis of arthritis and pioglitazone treatment with a change in any inflammation index as an outcome. Of the 1169 records initially identified following the selection process, two animal studies and four clinical studies were included in the review. Improvements from the baseline were observed in each treatment group for each inflammation indicator. The results of the animal studies on the temporomandibular joints and on patients with rheumatoid and psoriatic arthritis indicate that the drug in question may have potential to treat arthritis, including within the temporomandibular joint.

The action of TZDs is based on the activation of peroxisome proliferator-activated receptors (PPARs), which reduce insulin resistance, modify adipocyte differentiation, inhibit VEGF-induced angiogenesis, decrease leptin levels, reduce levels of various interleukins, and increase adiponectin levels [10][11][12][13][14]. The side effects of TZD intake include water retention and reduction in bone mineral density resulting in increased fracture risk, particularly in women, which may overlap with natural postmenopausal endocrine changes [15][16][17]. The only approved use of TZDs is currently to treat type 2 diabetes mellitus [1][2][3]. Albeit, experimental studies have been carried out on the therapeutic effect of TZDs in the treatment of numerous other diseases, such as polycystic ovary syndrome, ovarian hyperstimulation syndrome, nonalcoholic steatohepatitis, autism, psoriasis, and arthritis [18][19][20][21][22]. Studies were selected in which animals received pioglitazone as arthritis treatment (Problem A). In the identified studies, arthritis was pharmacologically induced in each case, as highlighted in Table 1 for clarity. Clinical trials for the treatment of any other type of arthritis in patients of any age and gender were included as Problem B. Due to the unknown mechanism of action of TZDs on the joints and the attempt to determine whether the therapy is causal or symptomatic, it was not decided to limit it to specific disease entities. The clinical diagnosis of arthritis was required to be radiologically confirmed or coexist with psoriasis.
For both Problems, to be able to compare the effectiveness of treatments in differently designed studies, the initial and final values of indicators of the severity of tissue inflammation were required to be available in the content of the report. All kinds of control groups and studies without a control group were allowed. The time frame for the included publications was not limited, but they were required to be available in English. Only research with published results was accepted, regardless of study design.

Information Sources and Search
A systematic search of medical databases was carried out based on 11 search engines: ACM Digital, BASE, EbscoHost, Embase, Ovid, ProQuest, PubMed, Scopus, Virtual Health Library, Web of Science, and Wiley Online Library [93][94][95][96][97][98][99][100][101][102][103]. One engine planned to be used, ScienceDirect, was omitted due to the lack of support for a query of the required complexity [104]. All final searches were made on 11 April 2022, under the same strategy: (arthritis OR osteoarthritis OR polyarthritis OR rheumatism OR rheumatic OR rheumatoid OR gout) AND (thiazolidinedione OR thiazolidinediones OR TZD OR glitazone OR glitazones OR pioglitazone OR actos OR rosiglitazone OR avandia OR lobeglitazone OR duvie OR ciglitazone OR darglitazone OR englitazone OR netoglitazone OR rivoglitazone OR troglitazone OR rezulin OR balaglitazone OR drf-2593 OR as-605240).
The list of individual search engines and queries tailored to the specificity of these tools is presented in Table A1.

Selection of Sources of Evidence, Data Charting Process, and Critical Appraisal of Individual Sources of Evidence
The identified records were processed by two authors (K.C. and M.C.) using the Rayyan tool (Qatar Computing Research Institute, Doha, Qatar and Rayyan Systems, Cambridge, MA, USA). After manual deduplication, the screening was carried out, and the convergence of judges' assessments at this stage was expressed by Cohen's kappa coefficient. In case of discrepancy in decisions, a given record was qualified for full-text evaluation on a par with those unanimously included. Further analysis of the full texts led to the final decision on each of the reports (K.C and M.C). Data were extracted from the content of qualified publications without the use of automation tools independently by two authors (N.T. and M.C.). The risk of bias was assessed for the included studies according to the RoB2 and ROBINS-I questionnaires for randomized and non-randomized trials, respectively.

Data Items
The following elements characterizing reports and study groups have been extracted: (1) First author; (2) Number of patients; (3) Diagnosis; (4) Daily dose, mg; (5) Treatment duration; (6) Inflammation indicator; (7) Initial value (100%) of this indicator; (8) Final value of this indicator; (9) Initial indicator value (100%) for control group; (10) Final indicator value for control group. If the exact number of days is not specified, it is assumed that one month consists of four weeks. A visual analog scale (VAS) with a different numerical range was converted proportionally to values from 0 to 10. The C-reactive protein concentrations were converted to mg/L.

Statistical Analysis (Synthesis of Results)
The final value of each indicator as a percentage of its original value was used as a measure of effect. It was calculated according to the formula developed for this synthesis: where e is the effectiveness of the therapy, f is the final value of the indicator for which the effect is measured, and i is the initial value of this indicator. The variables mentioned above were fully synthesized in the table. The values of the test probability p < 0.05 were considered statistically significant. The selected data are discussed in the text and presented in graphic form. Statistical analysis and visualization were performed using Microsoft Office software (Microsoft, Redmond, WA, USA).

Selection of Sources of Evidence
Searching medical databases using the above-mentioned search engines gave a total of 1169 records. Of these, the automation tool identified 683 potentially duplicate entries, of which 515 were manually deleted because of confirmation of duplication, leaving 168. After the deduplication process was completed, 654 records were screened. The convergence of the evaluations of two jurors at the screening stage was 99.54% (Cohen's κ = 0.91). In total, 375 abstracts were rejected due to inadequacy problems, including diagnoses other than the predetermined criteria, and cell line studies. Due to the wrong interventions and outcomes, 11 and five reports, respectively, were excluded. The wrong type of publication was found in 265 cases, including review articles and case reports. A foreign language resulted in the abandonment of the full-text evaluation of 27 items. Twenty reports were qualified for full-text evaluation, and six of them were finally included in the synthesis. The entire selection process is shown in Figure 1. The detailed numbers of records identified by each search engine are given in Table A2. The records rejected during full-text evaluation are summarized in Table A3. 0.91). In total, 375 abstracts were rejected due to inadequacy problems, including diagnoses other than the predetermined criteria, and cell line studies. Due to the wrong interventions and outcomes, 11 and five reports, respectively, were excluded. The wrong type of publication was found in 265 cases, including review articles and case reports. A foreign language resulted in the abandonment of the full-text evaluation of 27 items. Twenty reports were qualified for full-text evaluation, and six of them were finally included in the synthesis. The entire selection process is shown in Figure 1. The detailed numbers of records identified by each search engine are given in Table A2. The records rejected during full-text evaluation are summarized in Table A3.

Characteristics of Sources of Evidence, Critical Appraisal within Sources of Evidence and Results of Individual Sources of Evidence
In total, 27 tests on various indicators of inflammation were identified in six reports. The results obtained for six study groups and six control groups are presented below ( Table 2).

Characteristics of Sources of Evidence, Critical Appraisal within Sources of Evidence and Results of Individual Sources of Evidence
In total, 27 tests on various indicators of inflammation were identified in six reports. The results obtained for six study groups and six control groups are presented below (Table 2).

Animal Studies
Two of the reports discussed the effect of pioglitazone therapy in reducing the severity of TMJ arthritis in rabbits and mice [49,50]. The study by Shiojiri et al. did not specify the number of animals in the study group [50]. In both reports, due to the specificity of the inflammation indicators, the initial values were not known, and the percentage evaluation of the effectiveness of the treatment was calculated on the basis of the final values in the study and control groups [49,50]. The number of cartilage cell layers in the study by Kałużyński et al. is the only indicator for which values greater than 100% mean therapeutic success [49]. For the remaining papers (including clinical trials), the lower the percentage value is in the last column of the table (Final value as a%), the better the treatment effect was. TMJ-temporomandibular joint; TJC-tender joint count; SJC-swollen joint count; CRP-C-reactive protein level; ESR-erythrocyte sedimentation rate; DAS28-disease activity score for 28 joints; VAS-patient's assessment of pain using visual analog scale; IL-6-interleukin 6 level; TNF-alpha-tumor necrosis factor-alpha level; *-approximate value; N/S-not specified.

Clinical Trials
Another four reports concerned the treatment of a total of 172 patients suffering from systemic arthritis. The reports of both Shahin et al. and Bongartz et al. were ranked with a moderate risk of bias. This was due to their allocation of patients to treatment groups based on the severity of their baseline condition in the first study and with a lack of any blinding in the latter. Papers by Ormseth et al. and Marder at al. were judged to raise some concerns, but the main cause of this verdict was their missing data, despite the proper study design. In the study by Shahin et al., rheumatoid arthritis coexisted with diabetes mellitus; in Bongartz et al.'s study, patients were diagnosed with psoriatic arthritis; and in the other two clinical studies, rheumatoid arthritis was treated [65,67,105,106]. Differences in the daily doses and the number of days of treatment resulted in a difference in the total drug dose. The latter ranged from 2520 to 5040 mg [65,67,105,106]. Comparison of the baseline and end values of the arthritis indicators in the control groups using the Student's two-sided paired T-test showed that they were not statistically significant (p = 0.553). The same test showed a statistical significance between the initial and final values of the severity of inflammatory mediators (p = 0.007). The mean improvement for all tests in all study groups was 31%.

Synthesis of Results of Clinical Trials
The IL-6 and TNF-alpha rates were only measured in the study by Ormseth et al. [67]. The remaining markers of inflammation were repeated in the clinical reports that were analyzed, which allowed a graphical representation of the change in their rates (Figures 2-6) [65,67,105,106].
nal values in the study and control groups [49,50]. The number of cartilage cell layers in the study by Kałużyński et al. is the only indicator for which values greater than 100% mean therapeutic success [49]. For the remaining papers (including clinical trials), the lower the percentage value is in the last column of the table (Final value as a%), the better the treatment effect was.

Clinical Trials
Another four reports concerned the treatment of a total of 172 patients suffering from systemic arthritis. The reports of both Shahin et al. and Bongartz et al. were ranked with a moderate risk of bias. This was due to their allocation of patients to treatment groups based on the severity of their baseline condition in the first study and with a lack of any blinding in the latter. Papers by Ormseth et al. and Marder at al. were judged to raise some concerns, but the main cause of this verdict was their missing data, despite the proper study design. In the study by Shahin et al., rheumatoid arthritis coexisted with diabetes mellitus; in Bongartz et al.'s study, patients were diagnosed with psoriatic arthritis; and in the other two clinical studies, rheumatoid arthritis was treated [65,67,105,106]. Differences in the daily doses and the number of days of treatment resulted in a difference in the total drug dose. The latter ranged from 2520 to 5040 mg [65,67,105,106]. Comparison of the baseline and end values of the arthritis indicators in the control groups using the Student's two-sided paired T-test showed that they were not statistically significant (p = 0.553). The same test showed a statistical significance between the initial and final values of the severity of inflammatory mediators (p = 0.007). The mean improvement for all tests in all study groups was 31%.

Synthesis of Results of Clinical Trials
The IL-6 and TNF-alpha rates were only measured in the study by Ormseth et al. [67]. The remaining markers of inflammation were repeated in the clinical reports that were analyzed, which allowed a graphical representation of the change in their rates (Figures 2-6)
The results of clinical studies show a decrease in arthritis markers with pioglitazone therapy [65,67,105,106]. No TZDs other than pioglitazone were used in the included studies [65,67,105,106]. The change of TJC and SJC values over the course of the treatment in Shahin et al.'s and Bongartz et al.'s studies is promising [65,105]. The effect in these domains was much less pronounced than in the study by Ormseth et al., although the diagnosis and total dose were consistent with those of Shahin et al. [65,67]. In both studies, patients used pioglitazone as an add-on to their baseline therapy [65,67]. There was a clear difference in the length of therapy, with the study by Shahin et al. lasting 12 weeks and Ormseth et al.'s study lasting 8 weeks [65,67]. Therefore, it should be carefully assumed that a shorter administration time, despite the higher single doses, may give a worse anti-inflammatory effect [65,67]. Similar relationships were observed for the CRP index [65,67,105]. The results of the Shahin et al. and Ormseth et al. studies again proved to be in favor of the test group in the first report [65,67]. CRP was the only common indicator for all four studies [65,67,105,106]. The weakest improvement in the CRP domain was reported in the article by Marder et al., which could not be explained by the test conditions [106]. Studies of ESR and VAS pain ratios are difficult to interpret due to the small amount of inconsistent data [65,67,105].

Limitations
The limitations of the evidence in this scoping review are the heterogeneity of the studies and their concentration on pioglitazone, and the almost complete omission of rosiglitazone. Moreover, no clinical trials of TMJ arthritis have been identified, and only two animal-based studies are known.

Conclusions
Improvements from the baseline were observed in each treatment group for each inflammation indicator due to pioglitazone therapy. In relation to the control groups, better results were achieved in all combinations, except for SJC, TNF-alpha level, and DAS28-ESR, in the study of Ormseth et al. The results of animal studies on the temporomandibular joints and on patients with rheumatoid and psoriatic arthritis indicate that the drug in question may have potential to treat arthritis, including within the temporomandibular joint. Therefore, the development and conduct of clinical trials in collaboration with rheumatologists and maxillofacial surgeons in patients receiving pioglitazone as a treatment for hyperglycemia may be a promising direction for further research. Replacing treatments such as splint therapy and intra-articular injections with oral drugs also seems theoretically possible in the future, including for non-diabetic patients, but certainly cannot be considered in the current state of knowledge.

Conflicts of Interest:
The authors declare no conflict of interest.    [105]. Duplicate record Cuzzocrea et al. Rosiglitazone a ligands of the peroxisome proliferator-activated receptor-γ (PPAR-γ) reduce the evolution of murine type II collagen-induced arthritis [107].

Appendix A. For Materials and Methods
Animal non-TMJs study Xu et al. Implication for thiazolidinediones (TZDs) as novel potential anti-inflammatory drugs [108]. Review article Stojanovska et al. The anti-atherogenic effects of thiazolidinediones [109].
Animal non-TMJs study Inokuchi et al. Effects of benzbromarone and allopurinol on adiponectin in vivo and in vitro [11].
Wrong drug Fitzpatrick et al. The effects of rosiglitazone on bone by multiple image modalities in postmenopausal women with type 2 diabetes mellitus [111].
Conference abstract Ormseth et al. Peroxisome proliferator-activated receptor gamma agonist treatment for rheumatoid arthritis: A proof-of-concept randomized controlled trial [112].

Conference abstract
Ormseth et al. Reversing vascular dysfunction in rheumatoid arthritis: improved augmentation index but not endothelial function with peroxisome proliferator-activated receptor γ agonist therapy [113].

No inflammation indicators
Kim et al. Changes in bone mineral density in patients with type 2 diabetes treated with lobeglitazone, a novel thiazolidinedione, over 52 weeks: A multicenter, randomized, double-blind, placebo controlled trial [114].  [118]. Animal non-TMJs study