Prediabetes/diabetes screening strategy at the periodontal clinic

Abstract Objective The aim of the study was to propose an efficient chairside clinical strategy for the identification of undiagnosed hyperglycaemia in periodontal clinics. Material and methods Α chairside system was used for assessment of glycated hemoglobin 1c (HbA1c) and active Matrix Metalloproteinase‐8 levels (aMMP‐8) were analyzed by immunotest in patients (n = 150) who fulfilled the criteria for screening of the Centers for Disease Control and Prevention. Full‐mouth periodontal parameters were assessed and various data such as Body Mass Index (BMI), smoking and education were recorded. Results Thirty‐one patients out of 150 tested were found with unknown hyperglycaemia (20.7%). Regarding sex, education, parent with diabetes, normal BMI, smoking, age ≥45 years and prior testing for diabetes, no differences were observed between subjects displaying HbA1c < 5.7 and ≥5.7% (Pearson's Chi‐square test, p > .05). Subgroups differed regarding BMI (kg/m2), tooth count, percentages of 4 and 5 mm pockets (Mann–Whitney and z‐test, p < .05). The diagnostic performance for HbA1c ≥5.7 was tested by Receiving Operator Characteristic curves and Areas Under the Curve (AUC) for the following: age ≥ 45 years and BMI (AUC 0.651, p = .010), the above and aMMP‐8 (AUC 0.660, p = .006), age ≥ 45 years, BMI and Stage of Periodontitis (AUC 0.711, p < .001) and age ≥ 45 years, BMI, aMMP‐8 and stage of periodontitis (AUC 0.713, p < .001). Conclusions Findings of the study suggest that the combination of stage of periodontitis, increasing age, BMI and aMMP‐8, without chairside HbA1c assessment appears to be a viable screening strategy for referring dental patients for testing for prediabetes/diabetes.

asymptomatic during the initial stage of the disease. The global percentage of undiagnosed diabetes is alarming, and estimated to be 212.4 millions of adults in 2017, mainly from low and middle income countries (Cho et al., 2018). Prediabetes, defined as hyperglycaemia which is below the pathologic threshold but very close to it [Glycated hemoglobin 1c (HbA1c) 5.7-6.4%, and/or fasting plasma glucose (FPG) 100 mg/dL to 125 mg/dL] almost always precedes type 2 diabetes (C.D.C., 2017). However, as it has been solidly shown by randomized controlled clinical trials, lifestyle interventions are effective in preventing the progression of prediabetes to diabetes (Baker, Simpson, Lloyd, Bauman, & Singh, 2011;Howells, Musaddaq, McKay, & Majeed, 2016).
The aim of the present study is to propose a chairside point-ofcare (PoC) clinical strategy applied in patients attending periodontal clinics for the identification of undiagnosed hyperglycaemia.

| Subject sample
The minimally required sample size for identifying subjects with undiagnosed diabetes (n = 139) was calculated according to the estimated percentage of undiagnosed diabetes in Europe (10%) and by applying the relevant statistical equation as described before (Mataftsi, Koukos, & Sakellari, 2019

| Clinical procedures
In subjects who fulfilled the CDC criteria for developing diabetes type 2, the Cobas® b101 (Roche Diagnostics, Hoffmann La Roche, Mannheim, Germany) in vitro diagnostic test system for determination of HbA1c levels was applied. The system determines the amount of HbA1c in human capillary blood by photometric transmission measurement. The method has been standardized against the IFCC (International Federation of Clinical Chemists) reference method (Zhang et al., 2018). This diagnostic test also provides, at the same time, values of free glucose, when HbA1c is above 4.9%.
All subjects identified with hyperglycaemia (HbA1c ≥ 5.7%) were strongly advised to contact their physician for further consultation, and laboratory tests. Participants were also asked about the ease and convenience of the procedure and whether they would repeat it at the dental clinic.   (Table 2).
Firth's bias-reduced logistic regression was used for exploring the association between the active MMP-8 (aMMP-8) point-of-care test (ORALyzer®) and prediabetes, because there was a low prevalence for many variables. This leads to so called complete separation, which is a statistical challenge occurring when the dependent variable separates one (or more than one) variable completely. As a result, the estimates for the independent variables cannot be obtained, as the maximum likelihood does not exist. A common approach to this problem is to use Firth's method with a penalized maximum likelihood estimation (Heinze & Schemper, 2002). Receiver Operating Characteristic (ROC) curves were created using logistic regression models calculated by Firth's method and their predicted probabilities. The ROC curves were analyzed by Youden's index (Youden, 1950) to find efficient cut-off points for the models. Based on this cut-off and prediabetes prevalence, models were compared according to their performance in prediabetes diagnostics (Supplementary Table 2).
A two-sided p value <.05 was considered statistically significant in this study.
Patients characteristics according to HbA1c levels and periodontal condition are presented in Table 1 (Mataftsi et al., 2019).
In addition, subjects with HbA1c ≥ 5.7% displayed higher proportions of sites with clinical attachment loss >5 mm. These findings suggested that periodontal patients, especially those with bigger than normal BMI and waist circumference, are a target group worth screening for diabetes at the dental clinic (Mataftsi et al., 2019).
In the present study, participants were recruited irrelevant of periodontal condition and the percentage of not previously known hyperglycaemia was 20.7%. Both percentages are in agreement with the estimates for prediabetes in Greece as reported in studies from the medical field.
The recent consensus report of the joint workshop of the International Diabetes Federation and the European Federation of Periodontology reports that dentists dealing with patients without a diagnosis of diabetes are encouraged to apply screening methods and assess their risk for having diabetes, in order to refer to a physician for further testing identified subjects . In fact, the importance of validated questionnaires has been shown in a number of studies and they can be used with reasonable accuracy for prediabetes/diabetes screening (Bang et al., 2009;Herman, Smith, Thompson, Engelgau, & Aubert, 1995;Poltavskiy, Kim, & Bang, 2016;Rolka et al., 2001). This approach is certainly low-cost and therefore suitable for large-scale assessments both in clinical and community settings especially in low income countries. However, albeit this well established approach as shown in the current study according to model 1 (Figure 1), it is suggested that in the absence of chairside assessment of glycated hemoglobin A1c, an aMMP-8 chairside test could act-apart from periodontal inflammation-as a surrogate marker in order to refer patients for further evaluation by their physicians. This fact can contribute to the overall worldwide effort to limit the "pandemic" of diabetes type 2 along with several studies investigating the possible participation of dental practitioners to alert patients, by referring to medical practitioners and resulting to an early diagnosis and/or treatment.
Although these studies vary significantly regarding their design, sample size, age, racial and ethnic background of participants tested, they all conclude that it is feasible to screen dental patients for diabetes at the occasion of the dental visit (opportunistic screening). Noteworthy, that it has been reported that patients tend to visit their dentist on a more regular basis compared to their physician (Glick & Greenberg, 2005 .D.A., 2020) and the possibility of effects of ethnicity or hemoglobin variants (Barry et al., 2017).
According to the findings and in agreement with the literature subjects with HbA1c > 5.7 exhibited statistically significant differences in terms of clinical parameters of periodontal disease (Tables 1 and 2) thus underlying the contribution of hyperglycaemia to inflammation of periodontal tissues. This fact was also shown, when Periodontitis stage according to the 2018 classification (Papapanou et al., 2018) Figure 1). In the case of the POC aMMP-8 test, albeit its low cost and convenience, it is surely not intended for massively screening populations, but application at the dental office, especially for periodontitis patients as shown in Figure 1, can strengthen the reasons for a dentist to strongly recommend to a patient to get further checked by a physician and receive recommended instructions/treatment. It should be mentioned that this quantitative Point-of-Care test is commercially available and in use for online and real-time diagnosis and treatment monitoring of periodontitis (Alassiri et al., 2018;Grigoriadis et al., 2019;Johnson et al., 2016;Lorenz et al., 2017;Nwhator et al., 2014;Räisänen et al., 2018;Räisänen et al., 2019;Raivisto, Sorsa, Räisänen, et al., 2020;Sorsa et al., 2017), while assessment for HbA1c as applied in the present and other studies, is not yet easily feasible for a dental practice. A recent systematic review has shown that the dental workforce can be beneficially engaged in screening for prediabetes/diabetes but further clinical trials are required in order to optimize risk assessment protocols and strategies (Yonel et al., 2020).
Besides, the incorporation of validated biomarkers will improve diagnostic accuracy and assessment of stage and grade of the new Periodontitis classification system by Tonetti et al (Tonetti, Greenwell, & Kornman, 2018). and recent data have shown that the aMMP-8 mouth-rinse test can offer this possibility . This fact, is corroborated by findings of the present study, since, as shown in Table 2, the subgroup of participants with aMMP-8 above 20 ng/mL displayed statistically significant differences in clinical parameters of periodontal disease compared to subjects with ORALyzer® values below this threshold.
Taken collectively, findings of the present study suggest that the combination of periodontitis, increasing age, BMI and aMMP-8, when the use of chairside methods of HbA1c assessment is not available appears to be a viable screening strategy for correctly referring dental patients for further testing for prediabetes/diabetes by their physicians.

| CONCLUSIONS
In line with the previously published relevant reports, this study provides further supporting and extending evidence that the periodontal clinic is ideal for opportunistic screening for prediabetes/diabetes. Utilization of point-of care technology, that is, mouth-rinse aMMP-8 PoC assay enhances the ability of practitioners to contribute to the global effort for early diagnosis of prediabetes and/or T2 diabetes mellitus. in the collection, analyses or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.