Ki-67 expression as a diagnostic biomarker in odontogenic cysts and tumors: A systematic review and meta-analysis

Ki-67 is a marker of cell proliferation, used as an important diagnostic marker in the pathologic differentiation of various lesions. It is also relevant for developing targeted molecular therapies. We carried out a systematic review to assess the Ki-67 labeling index (LI) in odontogenic cysts and tumors. Databases were searched, including PubMed (MEDLINE), Scopus, CINHAL, PsycoInfo, the Cochrane Library, and Proquest. The meta-analysis was carried out based on the data of 608 lesions. When a 5% cut-off point was set, ki-67 LI of all benign odontogenic tumors dropped below this point. All the malignant tumors demonstrated an LI of over 15.3%; a significantly higher Ki-67 LI in malignant odontogenic lesions (17.59±2.80) was observed. Among benign tumors, the largest and the smallest Ki-67 LIs were seen in ameloblastoma (4.39±0.47) and adenomatoid odontogenic tumor (0.91±1.71). The mean values of Ki-67 LI in tumors and cysts were 4.23 (0.38) and 1.04 (0.07), respectively. Among odontogenic cysts, the highest Ki-67 LI was found in odontogenic keratocyst (OKC) (3.58±0.51), and the lowest in the radicular cyst (1.29±0.62%). Ki-67 LIs in all odontogenic cysts were <3%, except for OKC. This controversial lesion seems to have a profile more similar to a tumor, and a treatment plan similar to tumors might be suggested. We found that odontogenic lesions have diverse proliferative activities that help differentiate between various lesions and suggest therapeutic plans.


Introduction
Odontogenic lesions are one of the most common pathologic entities in the jaw region. Odontogenic cysts arise from different types of odontogenic epithelial cell types. The most common odontogenic cyst is the dentigerous cyst, which accounts for 20% of all the mandibular epithelial cysts. 1 An odontogenic keratocyst (OKC) is classified as a dental lamina cyst with a tendency to relapse after surgery, and it exhibits relatively aggressive clinical behavior. OKCs have a different origin, progression course, and biological behavior from other odontogenic cysts, e.g., dentigerous or radicular cysts. 2 Relatively high mitotic activity and increased epithelial cell turnover rates have been proposed to account for the growth of OKCs. 3 In 2005, OKC was classified as a benign odontogenic tumor in the WHO classification; however, it is still debated whether OKC is a benign tumor or an odontogenic cyst. 4 Ameloblastoma, the most prevalent clinically significant odontogenic tumor, comprises 1% of all jaw cysts or tumors, and 10-45.2% of all odontogenic tumors. Ameloblastoma is a benign tumor with a tendency for local invasion and a high recurrence rate; it is divided into solid/multicystic, unicystic, and peripheral subtypes. 5 Unicystic ameloblastoma displays a lining of focal or fully ameloblastic epithelium upon histopathological examination. 6 Ameloblastic carcinoma is a malignant epithelial odontogenic tumor that might develop within an existing ameloblastoma. 7,8 Abnormal cell proliferation is an essential feature of tumorigenesis. Ki-67 is a 319-358 kDa protein, considered a reliable marker for the proportion of proliferating cells and to predict the lesion's behavior. 9,10 Ki-67 is involved in all the active stages of the cell cycle (G1, S, G2, and M phases), but is not present in the resting G0 phase. 11 The TUOMS P R E S S most common standard for evaluating the tumor for Ki-67 expression is to calculate the percentage of positive nuclear staining with an anti-Ki-67 antibody, which is described as the Ki-67 labeling index (LI). 12 Ki-67 expression has been used in the diagnosis of many different human tumors, e.g., breast cancer, meningioma, prostate cancer, non-Hodgkin lymphoma, soft tissue sarcoma, etc. 13 Ki-67 is a diagnostic molecular marker involved in the growth of cancer cells. It is often assayed during the histopathological examination and has also been used for developing therapeutic methods. 14 In the maxillofacial region, Ki-67 has been used for distinguishing between adenoid cystic carcinoma (ACC) and polymorphous lowgrade adenocarcinoma (PLGA), two histopathologically similar lesions, but with different biological behavior and clinical outcome. Based on previous studies, the mean Ki-67 LI for PLGA and ACC has been reported to be 1.24% and 3.71%, respectively. 15 The Ki-67 LI has also been used to distinguish between two other lesions, basal cell adenocarcinoma (BCAC) and basal cell adenoma. The mean Ki-67 LI values in BCAC and BCA were 4.4% and 2.1%, respectively. The authors suggested a Ki-67 LI cutoff point of 5% and observed that 91% of BCAC LI values were higher than this cut-off point. Accordingly, a high Ki67 LI value could be helpful in BCAC diagnosis. 16 Ki-67 is a diagnostic and prognostic marker based on many statistical and laboratory findings and is, therefore, a promising candidate to differentiate between various lesions and suggest the best treatment approach. 14 This study aimed to systematically review the role of Ki-67 LI in odontogenic cysts and tumors. To the best of our knowledge, this is the first review assessing Ki-67 LI in odontogenic lesions.

Materials and Methods
This systematic review was carried out based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines for reporting systematic reviews.

Search strategies
PubMed (MEDLINE), Scopus, CINHAL, PsycoInfo, the Cochrane Library, and Proquest were used as the search databases.

Study selection and eligibility criteria
The inclusion criteria were as follows: The following studies were excluded: non-human research, reviews, letters, editorials, expert opinions, articles written in a language other than English, studies that reported ranges instead of means ± SD, studies in which the value was not expressed as a percentage, articles of poor quality, studies that did not include qualitative staining and immunohistochemistry analysis, duplicated articles (risk of bias).

Quality assessment
Using the CASP (Critical Appraisal Skills Program) quality checklist, the included articles were independently assessed by two specialists in oral and maxillofacial pathology (M.K. and M.V.), and low-quality studies were excluded.

Data extraction
Two investigators (M.K. and M.J.) independently extracted the articles' information and summarized them in the standardized extraction form. Extracted data comprised: first author's name, year of publication, study type, sample size, mean values and standard deviations of Ki-67 expression, lesion type (tumor/cyst), and the lesion name. Endnote X5 Resource Management Software was used to organize titles and abstracts of the articles and identify duplicates.

Statistical analysis
The Comprehensive Meta-Analysis (CMA; Englewood, NJ, USA) software, version 2.0, was used for data analysis. Mean and standard deviation values of ki-67 expression for each study were extracted from the selected articles. Q statistics and I 2 were used to determine heterogeneity. A probability value of I 2 value >50% and P value<0.10 for Cochran's Q indicated significant heterogeneity. Based on heterogeneity analysis, fixed or random-effects models were used. Subgroup analysis was performed based on the type of lesion, the lesion's name, and the sample size. Statistical significance was set at P<0.05.

Search results
The initial search retrieved 1458 articles. Of 1458 papers, 859 studies remained to be assessed after the duplicates were removed. A total of449 articles were excluded after reviewing the titles and abstracts, and the remaining 150 full-text articles were separately reviewed for quality by two investigators. Figure 1 presents the flowchart for the literature search and article selection procedures.

Characteristics of studies
At this stage, 132 studies were removed because the inclusion and exclusion criteria were not met. A few studies were excluded because of the method they used to report the immunohistochemical results. For example, in one study by Ramadoss et al, 17 the mean score of Ki-67 was reported instead of Ki-67 LI or percentage. In the study by Otero et al, 18 the mean ± SD was not reported separately for each lesion, and only a broad comparison was made. Also, in an article by Sathi et al, 19 the results of immunohistochemical staining were not reported as mean ± SD, and therefore, it was removed from the study. Finally, 18 studies were included in this meta-analysis. Table 1 presents the descriptive characteristics and related data of the included studies.

Publication bias
An estimation of potential publication bias was carried out using the funnel plot. The asymmetry of the plot suggested publication bias (Figure 2). Egger linear regression test (P<0.001) showed statistically significant publication bias.

Discussion
The present study was a systematic review to assess Ki-67 LI levels in odontogenic cysts and tumors. The issue's significance was confirmed by the growth in the number of papers over the period searched. This study's findings suggested increased Ki-67 LI levels in odontogenic tumors compared to cysts and in malignancies compared to benign lesions. Although Ki-67 has increasingly been studied as a prognostic indicator in recent years, this marker's more critical diagnostic role was emphasized in the present study. Simple histopathological techniques are not always able to differentiate between these challenging lesions. Random-effects model analysis was carried out for this purpose. Significantly higher Ki-67 LI levels were noted in ameloblastic carcinoma (17.59 ± 2.80%). The mean value obtained was significantly different from that in benign odontogenic lesions, consistent with previous studies. As observed by Yanamoto et al, 24 the Ki-67 LI levels in ameloblastic carcinoma (12.2 ± 12.2%) were higher than ameloblastoma (4.2 ± 3.6%). The findings of Bello et al 25 provide additional support for the statistically significant difference in Ki-67 LI levels between ameloblastoma and ameloblastic carcinoma.
The same crucial diagnostic question has been explored in studies carried out on lesions in other anatomical areas, and the results are comparable to our findings. Faur et al 38 reported higher Ki-67 expression in the malignant component of carcinoma tissue in pleomorphic adenoma than the benign component. The results were obtained from 29 cases of salivary gland tumors and the Ki-67 LI range for benign tumors was 0.5-6%, but the range was 3.6%-50% for malignant tumors. Therefore, the estimation of the potential proliferation rate of salivary gland tumors and identifying a subgroup of patients with invasive tumors, are useful in diagnostic pathology. This approach can differentiate between benign and malignant  lesions and predict the prognosis of salivary gland tumors. In another study by Tadbir et al 39 on Ki-67 expression in salivary gland tumors, the Ki-67 LI levels in malignant tumors (10.74 ± 10.8%) were higher than benign tumors, such as pleomorphic adenoma (0.78±0.2%).
As shown in Table 3, there were significant correlations between Ki-67 LI expression and the biopsy volume in odontogenic tumors, except for the volume of 3 µm in odontogenic tumors. These findings might suggest that biopsy volumes of 4-5 µm would be more appropriate for assessing Ki-67 LI expression in odontogenic tumors.
Among odontogenic cysts, the highest Ki67-LI expression was seen in OKC (3.58 ± 0.51%), and the lowest was observed in radicular cysts (0.98 ± 0.47%). These findings might be helpful in the histopathological analysis of cysts, especially when inflammation is present.
The order of Ki-67 LI expression in odontogenic cysts was as follows: OKC (3.58±0.51%), COC (2.43±0.65%), dentigerous cysts (1.29±0.62%), and radicular cysts (0.98 ± 0.47%) in descending order. These findings suggest that the assessment of Ki-67 LI in OKC during pathological diagnosis might be useful for diagnostic and therapeutic purposes. 40 Tsuneki et al 34 reported that the expression level of Ki-67 in OKC was higher than that in dentigerous and radicular cysts. However, in contrast to our findings, Živković et al 37 found that Ki-67 expression in OKC was lower than both dentigerous and radicular cysts; this finding might be explained because the evaluation only concerned the suprabasal layer. On the other hand, in a study by Kuroyanagi et al 40  In the present systematic review, covering 118 cases of OKC, 92 cases of radicular cysts, 33 cases of calcified odontogenic cysts (COC), and 48 cases of dentigerous cysts, we found the highest Ki-67 LI expression in OKC and the lowest expression in radicular cysts (Table 3). Therefore, it can be concluded that Ki-67 is expressed in developmental odontogenic cysts to a greater extent than inflammatory cysts, such as radicular cysts (0.98 ± 0.47%). Among developmental odontogenic cysts, the highest expression of Ki-67 was seen in OKC (3.58 ± 0.51%), with the lowest expression in dentigerous cysts (1.29 ± 0.62%). Selvi et al 42 reported higher Ki-67 expression in OKC compared with inflammatory and developmental cysts.
According to the data summarized in Table 2 (Ki-67      LI expression in odontogenic tumors), when a cut-off point of 5% is set for Ki-67 LI levels, all the benign tumors drop below this point, while Ki-67 LI for ameloblastic carcinoma is much higher (17.5 ± 2.8%). This fact plays an important diagnostic role and could suggest therapeutic approaches. In Table 4 (Ki-67 LI values in odontogenic cysts), it can be seen that in all odontogenic cysts, Ki-67 LI was less than 3%, except for OKC, which is regarded as a controversial lesion that, according to these findings, seems to be more similar to a tumor than a benign cyst. Although OKC is more commonly classified as an odontogenic cyst, it demonstrates features that distinguish it from other odontogenic cysts. 29 In the present study, the search strategy was limited to articles published in English. Articles that might contain data with potential high quality, written in other languages, were not included because of difficulties with accurate medical translation. In I 2 studies with statistical values that were <50%, we used the fixed-effects model and the Mantel-Hansel method, but in other studies where the values were >50% or the P-value was <0.05, the random-effects model was the most reliable option to calculate the overall effect size. The included studies' heterogeneity might be caused by different methodological techniques, imperfect reporting of cases, absence of true case verification, or other uncertain reasons. Despite these limitations, we believe that crucial diagnostic information can be extracted from the meta-analysis since the results were based on the best available evidence.

Conclusion
In this systematic review, we found that when a cut-off point of 5% was set, ki-67 LI of all benign odontogenic tumors dropped below this point. All malignant tumors demonstrated an LI of >15.3%; a significantly higher Ki-67 LI in malignant odontogenic lesions (17.59±2.80) was observed. Among odontogenic tumors, the maximum and the minimum Ki-67 LIs were found in ameloblastoma (4.39±0.47) and AOT (0.91±1.71). Among odontogenic cysts, the highest Ki-67 LI was seen in OKC (3.58±0.51),