Editor: Maurizio Tonetti

ISSN 0303 – 6979 VOLUME 48, NUMBER 1, JANUARY 2021

Editor: Maurizio Tonetti Oﬃ cial journal of the European Federa on of Periodontology Founded by the Bri sh, Dutch, French, German, Scandinavian and Swiss Socie es of Periodontology wileyonlinelibrary.com/journal/jcpe

EDITOR-IN-CHIEF: EDITORIAL BOARD: P. Hujoel, Seattle, WA, USA D. W. Paquette, Chapel Hill, NC, USA Maurizio Tonetti P. Adriaens, Brussels, Belgium J. Hyman, Vienna, VA, USA G. Pini-Prato, Florence, Italy Journal of Clinical Periodontology J. Albandar, Philadelphia, PA, USA I. Ishikawa, Tokyo, Japan P. Preshaw, Newcastle upon Tyne, UK Editorial Office G. Armitage, San Francisco, CA, J. Jansen, Nijmegen, the Netherlands M. Ryder, San Francisco, CA, USA John Wiley & Sons Ltd USA P.-M. Jervøe-Storm, Bonn, Germany G. E. Salvi, Berne, Switzerland 9600 Garsington Road, Oxford D. Botticelli, Rimini, Italy L. J. Jin, Hong Kong SAR, China A. Schaefer, Kiel-Schleswig-Holstein, OX4 2DQ, UK P. Bouchard, Paris, France A. Kantarci, Boston, MA, USA Germany E-mail: [email protected] A. Braun, Marburg, Germany D. F. Kinane, Louisville, KY, USA D. A. Scott, Louisville, Kentucky, USA K. Buhlin, Huddinge, Sweden M. Kebschull, Bonn, Germany A. Sculean, Berne, Switzerland ASSOCIATE EDITORS: M. Christgau, Düsseldorf, Germany M. Klepp, Stavanger, Norway L. Shapira, Jerusalem, Israel T. Berglundh, Göteborg, Sweden P. Cortellini, Florence, Italy T. Kocher, Greifswald, Germany B. Stadlinger, Zürich, Switzerland I. Chapple, Birmingham, UK F. Cairo, Florence, Italy E. Lalla, New York, NY, USA A. Stavropoulos, Aarhus C, Denmark R. Demmer, New York, NY, USA G. Dahlen, Gothenburg, Sweden N. P. Lang, Berne, Switzerland D. Tatakis, Columbus, OH, USA G. Hajishengallis, Philadelphia, PA, USA P. Eickholz, Frankfurt, Germany G. J. Linden, Belfast, UK R. Teles, Boston, MA, USA S. Jepsen, Bonn, Germany D. Fine, Newark, NJ, USA B. G. Loos, Amsterdam, the Netherlands L. Trombelli, Ferrara, Italy M. Quirynen, Leuven, Belgium T. Flemmig, Seattle, WA, USA H. Meijer, Groningen, Netherlands Y.-K. Tu, Taipei, Taiwan M. Sanz, Madrid, Spain W. V. Giannobile, Ann Arbor, J. Meyle, Giessen, Germany U. van der Velden, Amsterdam, the F. Schwarz, Düsseldorf, Germany MI, USA B. Michalowicz, Minneopolis, MN, USA Netherlands P. Sharpe, London, UK P. Gjermo, Oslo, Norway A. Mombelli, Geneva, Switzerland F. Vignoletti, Madrid, Spain F. Graziani, Pisa, Italy S. Murakami, Osaka, Japan H. Wachtel, Munich, Germany STATISTICAL ADVISER: A. Guerrero, Malaga, Spain I. G. Needleman, London, UK H.-L. Wang, Ann Arbor, MI, USA J. C. Gunsolley, Richmond, VA, USA A. Gustafsson, Stockholm, Sweden L. Nibali, London, UK J. L. Wennström, Gothenburg, Sweden P. A. Heasman, Newcastle, UK M. Nunn, Boston, MA, USA U. M. E.Wikesjö, Martinez, GA, USA EDITOR EMERITUS: D. V. Herrera, Madrid, Spain T. Oates, San Antonio, TX, USA J. Lindhe, Gothenburg, Sweden P. Holmstrup, Copenhagen, Denmark R. M. Palmer, London, UK

The aim of the Journal of Clinical Periodontology is to provide the platform for exchange of scientific and clinical progress in the field of Aim and periodontology and allied disciplines, and to do so at the highest possible level. The Journal also aims to facilitate the application of new scientific knowledge to the daily practice of the concerned disciplines and addresses both practicing clinicians and academics. The Journal Scope is the official publication of the European Federation of Periodontology but wishes to retain its international scope. The Journal publishes original contributions of high scientific merit in the fields of periodontology and implant dentistry. Its scope encompasses the physiology and pathology of the periodontium, the tissue integration of dental implants, the biology and the modulation of periodontal and alveolar bone healing and regeneration, diagnosis, epidemiology, prevention and therapy of periodontal disease, the clinical aspects of tooth replacement with dental implants, and the comprehensive rehabilitation of the periodontal patient. Review articles by experts on new developments in basic and applied periodontal science and associated dental disciplines, advances in periodontal or implant techniques and procedures, and case reports which illustrate important new information are also welcome.

The Journal of Clinical Periodontology is available to members of the American Academy of Periodontology, the Australian Society of Periodontology and the Canadian Academy of Periodontology at a reduced rate. Please contact your society for further details. For submission instructions, subscription and all other information visit: wileyonlinelibrary. com/journal/jcpe. Disclaimer The Publisher and Editor cannot be held responsible for errors or any consequences arising from the use of information contained in this journal; the views and opinions expressed do not necessarily reflect those of the Publisher and Editor, neither does the publication of advertisements constitute any endorsement by the Publisher and Editor of the products advertised. Copyright © 2021 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. All rights reserved. No part of this publication may be reproduced, stored or transmitted in any form or by any means without the prior permission in writing from the copyright holder. Authorization to copy items for internal and personal use is granted by the copyright holder for libraries and other users registered with their local Reproduction Rights Organisation (RRO), e.g. Copyright Clearance Center (CCC), 222 Rosewood Drive, Danvers, MA 01923, USA (www.copyright.com), provided the appropriate fee is paid directly to the RRO. This consent does not extend to other kinds of copying such as copying for general distribution, for advertising or promotional purposes, for creating new collective works or for resale. Special requests should be addressed to: [email protected]. Information for subscribers The JOURNAL OF CLINICAL PERIODONTOLOGY (ISSN 0303-6979 [print], ISSN 1600-051X [online]) is published monthly. US mailing agent: Mercury Media Processing, LLC 1850 Elizabeth Avenue, Suite #C, Rahway, NJ 07065 USA. Periodical postage paid at Rahway, NJ. Postmaster: Send all address changes to the JOURNAL OF CLINICAL PERIODONTOLOGY, Wiley Periodicals LLC, C/O The Sheridan Press, PO Box 465, Hanover, PA 17331 USA. Institutional subscription prices for 2021 are: Print & Online (premium): €1694.00 (Europe), US$2237.00 (The Americas), US$2608.00 (Rest of World). Prices are exclusive of tax. Australian GST, Canadian GST/HST and European VAT will be applied at the appropriate rates. For more information on current tax rates, please go to wileyonlinelibrary. com/tax-vat. The institutional price includes online access to the current and all online back files to January 1st 2017, where available. For other pricing options, including access information and terms and conditions, please visit wileyonlinelibrary.com/access. Where the subscription price includes print issues and delivery is to the recipient’s address, delivery terms are Delivered at Place (DAP); the recipient is responsible for paying any import duty or taxes. Title to all issues transfers FOB our shipping point, freight prepaid. We will endeavour to fulfil claims for missing or damaged copies within six months of publication, within our reasonable discretion and subject to availability. Back issues Single issues from current and prior year volumes are available at the current single issue price from [email protected]. Earlier issues may be obtained from Periodicals Service Company, 351 Fairview Avenue – Ste 300, Hudson, NY 12534, USA. Tel: +1 518 822-9300, Fax: +1 518 822-9305, Email: [email protected]. Publisher The Journal of Clinical Periodontology is published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK. Tel: +44 (0) 1865 776868; Fax: +44 (0) 1865 714591. Journal Customer Services: For ordering information, claims and any enquiry concerning your journal subscription please go to wileycustomerhelp.com/ask or contact your nearest office: Americas: Email: [email protected]; Tel: +1 781 388 8598 or 1 800 835 6770 (Toll free in the USA and Canada) Europe, Middle East and Africa: Email: [email protected]; Tel: +44 (0) 1865 778315 Asia Pacific: Email: [email protected]; Tel: +65 6511 8000 Japan: For Japanese speaking support, Email: [email protected]; Tel: +65 6511 8010 or (toll free): 005 316 50 480. Visit our Online Customer Get-Help available in 6 languages at wileycustomerhelp.com. Production Editor: Jenifer Jimenez (email: [email protected]) Advertising and commercial reprints: [email protected] Wiley’s Corporate Citizenship initiative seeks to address the environmental, social, economic, and ethical challenges faced in our business and which are important to our diverse stakeholder groups. We have made a long-term commitment to standardize and improve our efforts around the world to reduce our carbon footprint. Follow our progress at www.wiley.com/go/citizenship Wiley Online Library: This journal is available online at Wiley Online Library. Visit wileyonlinelibrary.com/journal/jcpe to search the articles and register for table of contents e-mail alerts. Hinari: Access to this journal is available free online within institutions in the developing world through the HINARI initiative with the WHO. For information, visit www.healthinternetwork.org. Note to NIH Grantees: Pursuant to NIH mandate, Wiley will post the accepted version of contributions authored by NIH grant-holders to PubMed Central upon acceptance. This accepted version will be made publicly available 12 months after publication. For further information, see www.wiley.com/go/nihmandate. COS -Printed in Singapore by C.O.S. Printers Pte Ltd. ISSN 0303-6979 (print) ISSN 1600-051X (online) Journal of Clinical Periodontology Ć Vol. 48 No. 1 January 2021

Contents

Periodontal Diseases The associations between major dietary patterns and risk of 2 A. A. Alhassani, F. B. Hu, Y. Li, B. A. Rosner, W. C. Willett periodontitis and K. J. Joshipura Obesity as a risk factor for tooth loss over 5 years: A population-based 15 A. C. Vallim, E. J. Gaio, R. V. Oppermann, C. K. Rösing, cohort study J. M. Albandar, C. Susin and A. N. Haas Association of hyperglycaemia with periodontal status: Results of the 25 P. Tegelberg, T. Tervonen, M. Knuuttila, J. Jokelainen, Northern Finland Birth Cohort 1966 study S. Keinänen-Kiukaanniemi, J. Auvinen and P. Ylöstalo The association between metabolic syndrome and periodontitis in 38 E. Montero, A. Molina, M. Carasol, A. Fernández-Meseguer, Spain: Results from the WORALTH (Workers’ ORAL healTH) Study E. Calvo-Bonacho, M. Teresa García-Margallo, M. Sanz and D. Herrera A quantitative bias analysis to assess the impact of unmeasured 51 T. S. Alshihayb, E. A. Kaye, Y. Zhao, C. W. Leone and confounding on associations between diabetes and periodontitis B. Heaton Periodontal Therapy The impact of smoking on non-surgical periodontal therapy: A 61 J. Chang, H.-W. Meng, E. Lalla and C.-T. Lee systematic review and meta-analysis Peripheral T helper cell profiles during management of periodontitis 77 N. Medara, J. C. Lenzo, K. A. Walsh, N. M. O’Brien-Simpson, E. C. Reynolds and I. B. Darby Periodontal surgery using rhFGF-2 with deproteinized bovine bone 92 H. Aoki, T. Bizenjima, F. Seshima, M. Sato, D. Irokawa, mineral or rhFGF-2 alone: 2-year follow-up of a randomized controlled K. Yoshikawa, W. Yoshida, K. Imamura, D. Matsugami, trial Y. Kitamura, D. Kita, H. Sugito, S. Tomita and A. Saito Periodontal infrabony defects: Systematic review of healing by defect 101 L. Nibali, D. Sultan, C. Arena, G. Pelekos, G.-H. Lin and morphology following regenerative surgery M. Tonetti Incidence and progression of gingival recession over 4 years: A 115 F. S. Rios, R. S. A. Costa, T. P. Wagner, B. R. Christofoli, population-based longitudinal study J. Goergen, C. Izquierdo, J. J. Jardim, M. Maltz and A. N. Haas Implant Therapy Post-extraction dimensional changes: A systematic review and meta- 127 E. Couso-Queiruga, S. Stuhr, M. Tattan, L. Chambrone and analysis G. Avila-Ortiz Volumetric soft tissue alterations in the early healing phase after peri- 146 C. M. Schmitt, P. Brückbauer, K. A. Schlegel, implant soft tissue contour augmentation with a porcine collagen matrix M. Buchbender, W. Adler and R. E. Matta versus the autologous connective tissue graft: A controlled clinical trial Corrigendum Corrigendum 164

This journal is abstracted or indexed in: Abstracts on Hygiene and Communicable Diseases, Bot anical Pesticides Abstracts, CAB Abstracts, Chemical Abstracts, Chemical Industry Notes, Cinahl: Cumulative Index to Nursing & Allied Health Literature, Current Abstracts, Current Contents, Current Titles in Dentistry, Dental Abstracts, Excerpta Medica. Abstract Journals, Forestry Abstracts, Global Health, Horticultural Science Abstracts, Index Medicus, Index to Dental Literature, Index to Scientiﬁ c Reviews, Index Veterinarius, MED LINE, Nutrition Abstracts and Reviews, Review of Aro matic and Medicinal Plants, Review of Medical and Veterinary Mycology, Rural Development Abstracts, Science Citation Index, SciSearch, SCOPUS, Soybean Abstracts,Tropical Diseases This journal is available online.Visit wileyonlinelibrary. Bulletin,Veterinary Bulletin, Veterinary Science Database com/journal/jcpe to search the articles and register for table of contents e-mail alerts. Printed in Singapore Received: 6 August 2019 | Revised: 19 September 2020 | Accepted: 24 September 2020 DOI: 10.1111/jcpe.13380

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

The associations between major dietary patterns and risk of periodontitis

1Department of Nutrition, Harvard T. H. Chan School of Public Health, Boston, Abstract MA, USA Aim: To prospectively investigate the associations between major dietary patterns 2Department of Epidemiology, Harvard T. H. Chan School of Public Health, Boston, and incidence of periodontitis. MA, USA Methods: We included 34,940 men from the Health Professionals Follow-Up Study, 3 Channing Division of Network Medicine, free of periodontal disease and major illnesses at baseline. Detailed medical and Department of Medicine, Brigham and Women's Hospital and Harvard Medical dental history was collected through biennial mailed questionnaires, and dietary in- School, Boston, MA, USA formation was provided through quadrennial food frequency questionnaires. Using 4Department of Biostatistics, Harvard T. principal component analysis, we identified two major dietary patterns (“prudent” and H. Chan School of Public Health, Boston, MA, USA “Western”). We used Cox proportional hazard models to examine the associations 5 Center for Clinical Research and Health between the two dietary patterns and self-reported incidence of periodontitis over a Promotion, University of Puerto Rico Medical Sciences Campus, San Juan, PR, 24-year follow-up period. We investigated each pattern separately. USA Results: There was no overall association between Western or prudent dietary pat-

Correspondence terns and periodontitis. Among obese, however, the Western dietary pattern was Kaumudi J. Joshipura, Center for Clinical significantly associated with incident periodontitis. The hazard ratio for those in the Research and Health Promotion, Office A-107, University of Puerto Rico Medical highest quintile of Western diet versus those in the lowest (reference) was 1.83 (95% Sciences Campus, San Juan, PR 00936, confidence interval: 1.21–2.76). USA. Email: [email protected] Conclusions: There was no overall association between Western or prudent dietary patterns and periodontitis; however, in subgroups analysis, the Western diet was sig- Funding information National Institutes of Health, Grant/Award nificantly associated with higher periodontitis risk only among obese men, a finding Number: UM1 CA167552 that requires replication and biological explication.

KEYWORDS diet, dietary pattern, epidemiology, periodontal disease, periodontitis

1 | INTRODUCTION tooth loss and diet (Joshipura & Andriankaja, 2016). Only a few established risk factors of periodontal disease are modifiable. Thus, Periodontitis is one of the most common diseases in adults world- identification of additional modifiable risk factors is important for wide. In addition to the deleterious effect on oral health, peri- better understanding of periodontitis pathology and to enable bet- odontitis is linked to multiple systemic health conditions, including ter cost-effective public health efforts targeting periodontal disease cardiovascular disease and impaired glycaemic control (Otomo- (Joshipura & Andriankaja, 2016). Corgel et al., 2012). These associations could be driven by systemic The relationship between diet and periodontitis is poorly un- inflammation, or through the impact of periodontal disease on derstood. The premise that diet could affect periodontal health is

2 | wileyonlinelibrary.com/journal/jcpe J Clin Periodontol. 2021;48:2–14. ALHASSANI et al. | 3 scientifically supported by animal studies, a few human studies and a strong biological plausibility (Chapple, 2009; Dietrich & Hoffmann, Clinical Relevance 2004; Fujita & Maki, 2015; Kaye, 2012; Muluke et al., 2016). However, Scientific rationale for the study: The relationship between the vast majority of published human studies were cross-sectional, diet and periodontitis is poorly understood. There is a pau- and most studies focused on nutrients directly relevant to epithelial city of literature on the long-term impact of dietary pat- and mineralized connective tissue health of the periodontium, that is terns on periodontal health. vitamins C, A and D and minerals such as calcium (Kaye, 2012). A few Principal findings: There was no overall association between studies investigated other nutrients, such as dietary fibre, omega-3 the investigated major dietary patterns and periodontitis. fatty acids or specific food groups (Iwasaki et al., 2010; Merchant Only among obese individuals, a Western diet character- et al., 2006; Nielsen et al., 2016; Schwartz et al., 2012; Yoshihara et al., ized by a higher intake of processed meat, red meat, but- 2009). Al-Zahrani et al. (2004) used the National Health and Nutrition ter, high-fat dairy products, eggs and refined grains was Examination Survey (NHANES) III to study the association between related to a higher risk of periodontal disease. diet quality using the original healthy eating index (HEI) and periodon- Practical implications: Diet could be a modifiable risk factor tal disease and found a positive association between poor dietary of periodontitis only among obese individuals if corrobo- quality and the degree of calculus deposition on teeth. In another rated in further research in different populations. study, the same authors found an inverse association between three health-enhancing behaviours: maintaining normal weight, engaging in recommended level of exercise and consumption of a high-quality diet, and the prevalence of periodontal disease (Al-Zahrani et al., 2005b). Dietary pattern analysis is an approach in nutritional epidemi- history in addition to lifestyle behaviour and body measurements ology that allows studying the overall role of diet in disease risk was collected through biennial questionnaires. Participants provided instead of limiting focus to certain foods or nutrients. Since diet information about their diet via semi-quantitative food frequency is a complex exposure, pattern analysis approach provides a plat- questionnaires (FFQs) every 4 years. The reproducibility and valid- form that better accounts for interaction and correlation between ity of the FFQ in measuring diet for this cohort have been reported nutrients, and allows identification of cumulative effects of foods previously (Rimm et al., 1992; Willett, 2012). and nutrients (Hu, 2002). Principal component analysis (PCA) is one We excluded participants who had periodontitis at baseline of the approaches for developing dietary pattern, and it has been (n = 8333) and those who were edentulous (n = 485), because they conducted in the Health Professionals Follow-Up Study (HPFS) and were not at risk of incident periodontal disease. Participants who showed that most of the variability in diet in this cohort could be ex- reported myocardial infarction (n = 1486), coronary artery surgery plained by two factors, named prudent and Western dietary patterns (n = 671), diabetes (n = 809) or cancer (n = 1317) at baseline were (Hu et al., 1999). The Western pattern is high in processed meat, excluded, due to the strong likelihood that those events could red meat, butter, high-fat dairy products, eggs and refined grains strongly modify dietary habits. We excluded participants who only while the prudent pattern is high in vegetables, fruit, legumes, whole responded to the baseline questionnaire (n = 3309), those who had grains, fish and poultry (Hu et al., 1999). Prudent and Western pat- missing periodontal information (n = 1117) and those with missing terns have been investigated in the medical literature. For instance, data on body mass index (BMI) (n = 902), physical activity (n = 128) a Western diet is associated with increased risk of diabetes, cancer or age (n = 36) at baseline. Participants who reported caloric intake recurrence and cardiovascular and total mortality, while a prudent outside the plausible range (800–4200 kcal/day), and those who left diet is inversely associated diabetes, cardiovascular and total mortal- 70 or more out of 131 FFQ items blank were excluded (n = 1033) ity (Ardisson Korat et al., 2014; Astrup et al., 2011; Fung et al., 2004; (Jimenez et al., 2012a, 2012b, 2014; Merchant et al., 2006; Pitiphat Heidemann et al., 2008; Kiefte-de Jong et al., 2017; Malik et al., et al., 2003). The final analysis included 34,940 men at baseline. 2012; Meyerhardt et al., 2007; Qi et al., 2009; van Dam et al., 2002). The study was approved by the institutional review boards of the The aim of the current study was to prospectively investigate the Brigham and Women's Hospital and the Harvard T.H. Chan School association between prudent and Western patterns and incidence of of Public Health. periodontal disease in the HPFS over 24 years of follow-up.

2.2 | Outcome assessment 2 | METHODS The main outcome of the study was incidence of periodontal disease, 2.1 | Study population defined as answering “yes” to the question “Have you been professionally diagnosed with periodontal disease with bone loss?” in the The HPFS consists of 51,529 male health professionals who com- biennial mailed questionnaire. Self-reported periodontal disease has pleted the baseline mailed questionnaire in 1986 when they were been evaluated in a subsample of the HPFS against bitewing radio- 40–75 years old. Detailed information about medical and dental graphs and showed good validity (Joshipura et al., 1996, 2002). 4 | ALHASSANI et al.

2.3 | Main exposures assessment into quintiles. Energy adjustment aids in controlling for extraneous errors in data (e.g. over-reporting or underreporting) (Willett, 2012). The main exposure for the study was PCA-derived dietary patterns (Hu et al., 1999). Briefly, for each cycle, the ~130 items in the FFQs were classified into ~40 predefined food groups based on nutrient 2.5 | Statistical analysis composition and on culinary use, and the PCA was applied using those food groups. The factors were rotated using varimax trans- Baseline descriptive statistics for the study cohort by quintiles of formation to develop uncorrelated factors with simpler structure the Western and prudent dietary patterns were calculated. We used and easier interpretability. Most of the dietary variability could be Cox proportional hazard models with age in months as the under- explained by the two dietary factors. The first was characterized by lying timescale. Person-time was calculated from the return of the a higher intake of vegetables, fruit, legumes, whole grains, fish and baseline questionnaire until the earliest of the following: incidence poultry and was named prudent dietary pattern. The second was of periodontal disease, mortality, last available response or end of characterized by a high intake of processed meat, red meat, but- follow-up (31 January 2010). The hazard ratio was calculated by ter, high-fat dairy products, eggs and refined grains and was named comparing the higher quintiles of each pattern to the lowest quintile. Western dietary pattern. On each FFQ cycle, a participant would To conduct the test for linear trend, we assigned the median value have a score for the prudent pattern and a score for the Western of each pattern's quintile and used these as a continuous variable. pattern. Scores were calculated by summing the standardized intake We handled missing exposure or covariates data during follow-up by of the food groups weighted by their factor loadings. To better rep- carrying forward values from the last cycle. resent the long-term dietary intake, and to minimize measurement In order to understand the potential role of adiposity in our error and within-person variability, the cumulative average of each study, we compared the models with and without inclusion of BMI to pattern score from baseline until start of each 2-year follow-up in- evaluate whether BMI is an important mediator or confounding vari- terval was calculated and divided into quintiles. The reproducibility able. We also did the analysis stratifying by updated BMI categories and validity of the PCA-derived dietary patterns have been assessed (18.5–24.9, 25–29.9 and ≥30). To explore the relationship within lev- in a subsample of the HPFS (Hu et al., 1999). The Pearson correla- els of the covariates, we stratified the analysis by occupation (den- tion between 2 FFQs approximately one year apart was 0.70 for the tist versus non-dentist), age (≥65 vs. <65), physical activity (above prudent pattern scores and 0.67 for the Western pattern scores (Hu the median versus below) and diabetes over the follow-up. To test et al., 1999). We investigated each pattern separately. We did the for the statistical significance of interaction, we created indicator analysis using the Western pattern as an exposure and then did the variables for being a dentist, updated binary physical activity level, analysis using the prudent pattern as an exposure. binary age, diabetes status and obesity status (BMI ≥ 30). We fitted adjusted models that included interaction terms between the continuous variables of dietary patterns (same variables used for linear 2.4 | Covariates assessment trends) and the created indicator variables. For smoking, instead of an indicator variable, we used continuous CSI value. We used a Wald The analysis was adjusted for important predictors of periodontal test to calculate the p-value of interaction. We assessed the propor- disease, that is age, smoking, BMI, physical activity and alcohol con- tional hazards assumption using a likelihood ratio test comparing the sumption (Albandar, 2005; Al-Zahrani et al., 2003, 2005a, 2005b; model with and without an interaction term between time period Jimenez et al., 2012a; Joshipura & Andriankaja, 2016; Pitiphat and the exposure. Analysis was performed using SAS for UNIX sta- et al., 2003). We controlled for smoking by using the Comprehensive tistical software (version 9.4; SAS Institute). Smoking Index (CSI), which takes into consideration the duration of smoking in years, smoking intensity, time since smoking cessation in years and a specific biological half-life of smoking effect on the dis- 3 | RESULTS ease (Leffondre et al., 2002; Pitiphat et al., 2003). BMI values were categorized as follows (<18.5, 18.5–24.9, 25–29.9 and ≥30 kg/m2); The number of new cases of periodontal disease was 3738 over updated measures every 2 years were used since Jimenez et al. pre- the 24-year follow-up (747,517 person-years). Table 1 shows the viously reported a positive association between updated BMI and distribution of age-adjusted baseline characteristics by quintiles periodontal disease (Jimenez et al., 2012a). For physical activity, we of Western and prudent dietary patterns. Participants with higher used the Metabolic Equivalent of Task (MET) to calculate MET hours scores of Western diet were less physically active, had fewer teeth (MET-hs) per week and categorized data into quintiles. We used the and were more likely to be current smokers at baseline. Men with updated measures for each follow-up. Incidence of diabetes during higher scores of prudent diet were more physically active, consumed the follow-up was collected from the biennial questionnaire. The less alcohol and were less likely to be currently smoking at baseline. cumulative averages of total alcohol and total caloric intake were Dentists in the cohort tended to have lower Western diet scores, estimated from the FFQs. We classified alcohol as 0, 0.1–4.9, 5–14.9, and higher prudent diet scores. Trends for the aforementioned char- 15–29 and ≥30 grams per day (g/day) and categorized caloric intake acteristics looked linear across quintiles of both patterns. ALHASSANI et al. | 5

The hazard ratio of periodontal disease in the highest quintile 1.21–2.76, p-value for trend <.01), whereas the association was null of Western diet in the age-adjusted model was 1.18 (95% CI: 1.07- in overweight and normal weight men (p-value for interaction = .03). 1.30, p-value for trend <.001; Table 2). This weak but statistically In a separate analysis, we adjusted for continuous BMI value in the significant association became non-significant in the multivariate models stratified by BMI categories and that did not affect the re- adjusted models. The prudent pattern was not associated with the sults (HR: 1.83, 95% CI: 1.21–2.76, p-value for trend <.01) (Table S1, risk of periodontal disease in our analysis. Adjusting for BMI in the model B). We also examined the joint associations between dietary models did not significantly change the association (model 3 versus patterns and BMI in relation to incidence of periodontitis (Figure 1). model 2). The associations were similar among subgroups defined The results from the joint classification were similar to the stratified by age, physical activity level, diabetes, smoking and profession, analysis, where obese individuals with the highest intake of Western for both Western and prudent dietary patterns. However, we de- diet had the highest risk of periodontal disease. tected a statistically significant effect modification by obesity for Our results among obese individuals were robust to multiple Western dietary pattern (Table 3). The risk of periodontal disease sensitivity analyses (Table S1). Due to the complex relationship be- for obese individuals in the highest quintile of Western pattern was tween diabetes and periodontal disease, it is reasonable to assume 83% higher than the risk in the lowest quintile (HR: 1.83, 95% CI: that diabetes could be in the causal pathway between diet and

TABLE 1 Age-standardized characteristics of the HPFS study population at baseline by quintiles of the baseline Western and prudent diet scores (1986)

Western Prudent

Diet pattern Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5

N (%) 6126 (17.5%) 7254 (20.8%) 7177 (20.5%) 6872 (19.7%) 7209 (20.6%) 6671 (19.1%) Age 53.9 ± 9.5 52.3 ± 9.4 51.0 ± 9.1 50.8 ± 9.2 52.4 ± 9.4 53.5 ± 9.5 White 93% 96% 97% 95% 95% 95% BMI (kg/m2) 24.9 ± 3.2 25.5 ± 3.4 26.0 ± 3.6 25.6 ± 3.3 25.4 ± 3.2 25.3 ± 3.5 Smoking Former 41% 40% 39% 39% 40% 41% Current 5% 8% 13% 15% 8% 5% CSI score 0.07 ± 0.43 0.10 ± 0.52 0.18 ± 0.70 0.21 ± 0.76 0.10 ± 0.51 0.07 ± 0.42 Alcohol gm/day 11.7 ± 16.4 10.9 ± 14.4 11.3 ± 15.7 14.4 ± 19.2 10.5 ± 14.2 9.8 ± 12.9 Physical Activity MET/Week 27.1 ± 35.5 21.5 ± 28.1 18.3 ± 26.4 16.2 ± 23.0 21.6 ± 29.7 29.0 ± 37.2 Dentist 66% 58% 45% 48% 58% 64% Number of teeth 25–32 90% 89% 85% 85% 89% 91% 17–24 8% 9% 11% 11% 9% 8% 11–16 1% 1% 2% 2% 1% 1% 1–10 1% 1% 1% 2% 1% 1% Foods groups (serving/day) Processed meat 0.08 ± 0.11 0.30 ± 0.22 0.81 ± 0.63 0.46 ± 0.52 0.37 ± 0.38 0.30 ± 0.38 Red meat 0.21 ± 0.16 0.57 ± 0.28 1.12 ± 0.54 0.67 ± 0.47 0.63 ± 0.45 0.56 ± 0.46 Butter 0.10 ± 0.23 0.25 ± 0.52 0.58 ± 1.03 0.32 ± 0.69 0.29 ± 0.63 0.26 ± 0.62 High-fat dairy 0.48 ± 0.47 0.91 ± 0.80 1.57 ± 1.44 0.98 ± 1.08 0.98 ± 0.97 0.95 ± 0.98 Eggs 0.15 ± 0.19 0.30 ± 0.31 0.58 ± 0.62 0.35 ± 0.46 0.33 ± 0.39 0.32 ± 0.41 Refined grains 0.72 ± 0.60 1.10 ± 0.85 1.96 ± 1.43 1.17 ± 1.06 1.23 ± 1.05 1.31 ± 1.11 Vegetables 3.18 ± 2.10 2.86 ± 1.67 3.07 ± 1.68 1.31 ± 0.50 2.66 ± 0.60 5.50 ± 1.99 Fruits 1.99 ± 1.65 1.53 ± 1.17 1.46 ± 1.09 0.78 ± 0.61 1.50 ± 0.90 2.69 ± 1.75 Legumes 0.47 ± 0.44 0.45 ± 0.34 0.55 ± 0.42 0.26 ± 0.17 0.44 ± 0.26 0.84 ± 0.57 Whole grains 1.62 ± 1.51 1.52 ± 1.36 1.61 ± 1.47 0.88 ± 0.86 1.56 ± 1.28 2.34 ± 1.84 Fish 0.49 ± 0.41 0.38 ± 0.29 0.34 ± 0.27 0.22 ± 0.17 0.38 ± 0.26 0.61 ± 0.45 Poultry 0.37 ± 0.28 0.34 ± 0.27 0.34 ± 0.26 0.23 ± 0.18 0.35 ± 0.24 0.48 ± 0.36

Note: Values are means ± SD or percentages and are standardized to the age distribution of the study population. 6 | ALHASSANI et al.

TABLE 2 Hazard ratios (95% CIs) of incidence of periodontitis according to quintiles of the Western and prudent diet scores

Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 p Trend

Western diet Cases/p-years 763/140,732 678/141,223 710/141,435 740/141,817 847/141,964 Model 1 1 0.93 (0.84–1.03) 0.98 (0.89–1.09) 1.03 (0.93–1.13) 1.18 (1.07–1.30) <.001 Model 2 1 0.91 (0.82–1.01) 0.95 (0.85–1.06) 0.97 (0.87–1.09) 1.07 (0.94–1.22) .13 Model 3 1 0.90 (0.81–1.00) 0.94 (0.84–1.05) 0.96 (0.85–1.08) 1.05 (0.92–1.19) .24 Prudent diet Cases/p-years 738/140,354 738/141,155 705/141,693 775/141,919 775/141,919 Model 1 1 0.99 (0.89–1.09) 0.93 (0.83–1.03) 1.00 (0.91–1.11) 0.98 (0.89–1.09) .93 Model 2 1 1.03 (0.93–1.14) 0.98 (0.88–1.09) 1.07 (0.96–1.19) 1.04 (0.93–1.16) 0.41 Model 3 1 1.03 (0.92–1.14) 0.98 (0.88–1.09) 1.07 (0.96–1.19) 1.04 (0.93–1.16) .41

Note: Model 1: age adjusted. Model 2: adjusted for age, energy intake (quintiles), smoking (CSI), physical activity (METs quintiles), alcohol (g/day: 0, 0.1–4.9, 5–14.9, 15–29, 30+), occupation (dentist vs. non-dentist), and race (White/Black/Asian/Other). Model 3: model 2 and adjusted for BMI (<18.5, 18.5–24.9, 25–29.9, 30+).

periodontal disease. Also, development of diabetes may significantly (e.g. whole grains) were associated with a lower risk of periodon- modify dietary habits. Hence, we conducted the analysis both with tal disease, their significance may have been diluted in the current and without adjusting for incident diabetes during follow-up and the analysis which can explain the null results of prudent diet. Our find- results were similar. In a separate analysis, we censored individuals ings regarding Western diet are in fair agreement with the limited when they reported diabetes diagnosis during follow-up, and the re- published work that studied some of the Western diet components. sults remained similar. Merchant et al. did not find an association between refined grains and periodontal disease (Merchant et al., 2006). Yoshihara et al. combined “fish, shellfish, meat, bean and eggs” and found no rela- 4 | DISCUSSION tionship with periodontitis. An inverse association between dairy consumption and periodontitis was reported in cross-sectional stud- We did not see an overall relation between prudent or Western ies (Adegboye et al., 2012, 2016; Al-Zahrani, 2006; Shimazaki et al., dietary patterns and risk of periodontal disease over the 24-year 2008), but the longitudinal study by Yoshihara et al. did not detect follow-up. In the subgroups analysis, we found a robust association a correlation between “milk and milk products” and periodontal dis- between Western diet and risk of periodontal disease only among ease (Yoshihara et al., 2009). obese individuals, which persisted after controlling for several con- There is no consensus on the case definition of periodontitis in founders and through several sensitivity analyses. epidemiological studies (Holtfreter et al., 2015). We defined incidence To our knowledge, this is the first study that related diet with of periodontitis as answering “yes” to the question “Have you been periodontitis using a data-driven posteriori approach, PCA, to iden- professionally diagnosed with periodontal disease with bone loss?”.It tify dietary patterns. Hence, the ability to compare our results to is inevitable that some degree of misclassification has occurred, which the current literature is limited to specific components of the two could have attenuated our results (Rothman et al., 2008). However, patterns. Using the HPFS, Merchant et al. found that men in the self-reported periodontal disease showed acceptable validity in a sub- highest quintile of whole-grain intake were 23% less likely to de- sample of the HPFS. Among the dentists in HPFS, the positive pre- velop periodontitis than those in the lowest quintile (95% CI: 11%, dictive value for self-reported periodontal disease was 0.76 and the 34%) (Merchant et al., 2006). Yoshihara et al. reported a negative negative predictive value was 0.74. Among non-dentists, those values correlation between intake of vegetables and incidence of peri- were 0.83 and 0.69, respectively, making the method an acceptable odontal disease, and a positive correlation with intake of cereals, proxy and a valid “endpoint” in this cohort (Joshipura et al., 1996). The nuts, seeds, sugar, sweeteners and confectioneries (Yoshihara et al., question wording in our cohort was reported to perform well when 2009). Schwartz et al. found that dietary fibre intake was associ- compared to other methods of self-reported periodontal disease ated with a reduced progression of periodontal disease among older (Blicher et al., 2005). Also, several associations with periodontal dis- men (Schwartz et al., 2012). Our pattern approach to assessing diet ease as an outcome have been established in the HPFS (Jimenez et al., is more analogous to the “real world” where people do not consume 2012a, 2012b, 2014; Pitiphat et al., 2003). Hence, we do not believe single nutrients, but rather eat meals with combinations of food and that the null results are primarily attributable to the outcome ascertain- nutrients that potentially interact synergistically or antagonistically ment method. However, patient-level clinical diagnosis of periodontitis (Hu, 2002). Hence, although some of the prudent diet components is almost never described as all or none and usually includes multiple ALHASSANI et al. | 7

‡ (Continues) .27 .32 (Obese vs. Non-obese) vs. (Obese .03 interaction p

† .84 .17 .50 .39 <.01 p Trend .58 .52 1.01 (0.84–1.21) 1.01 374/68,835 1.09 (0.91–1.32) 1.09 473/73,129 1.05 (0.86–1.28) 1.05 301/46,680 1.05 (0.88–1.25) 546/95,284 1.83 (1.21–2.76) 156/21,109 1.02 (0.83–1.24) 0.93 (0.76–1.14) 0.93 Quintile 5 Quintile 283/50,680 404/69,574 0.89 (0.75–1.06) 0.89 323/70,232 1.03 (0.88–1.22) 417/71,584 0.89 (0.75–1.07) 0.89 293/53,455 1.01 (0.87–1.19) 1.01 447/88,361 1.34 (0.91–1.99) 101/18,524 1.00 (0.84–1.19) 0.85 (0.71–1.02) 0.85 Quintile 4 Quintile 263/54,005 374/68,812 0.90 (0.77–1.06) 0.90 335/73,079 0.99 (0.85–1.16) 375/68,356 0.88 (0.75–1.04) 0.88 297/54,813 0.98 (0.85–1.14) 413/86,622 1.37 (0.94–1.98) 92/15,979 0.94 (0.80–1.11) 0.94 0.87 (0.73–1.02) 0.87 Quintile 3 Quintile 273/56,977 344/67,956 0.89 (0.77–1.03) 0.89 333/73,331 0.92 (0.79–1.07) 0.92 345/67,892 0.84 (0.72–0.99) 290/56,526 0.95 (0.82–1.09) 0.95 388/84,697 1.37 (0.95–1.96) 85/14,382 0.89 (0.76–1.04) 0.89 0.83 (0.71–0.98) 0.83 Quintile 2 Quintile 274/61,087 317/65,301 1 408/77,499 1 355/63,232 1 357/57,164 1 406/83,568 1 49/10,824 1 1 Quintile 1 Quintile 383/70,731 330/58,644 years years years years years years years HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) HR (95% CI) Cases/p- Cases/p- Diabetes No ≥Median Physical activityPhysical

‡ (Continues) .81 .28 .84 interaction p

† .50 .56 .46 .20 .37 .36 .17 p Trend 1.03 (0.87–1.22) 1.03 393/62,895 1.06 (0.86–1.30) 1.06 0.91 (0.75–1.11) 454/79,069 1.43 (0.72–2.87) 313/74,076 1.06 (0.87–1.30) 1.06 38/5,890 300/52,890 1.21 (0.82–1.79) 1.21 1.04 (0.91–1.18) 205/16,158 Quintile 5 Quintile 809/136,074 0.94 (0.80–1.09) 0.94 403/75,285 0.99 (0.81–1.20) 0.99 0.94 (0.79–1.12) 0.94 337/66,532 1.59 (0.88–2.89) 320/76,865 0.94 (0.78–1.13) 0.94 40/5,533 307/52849 1.05 (0.72–1.52) 0.94 (0.83–1.06) 0.94 120/12,062 Quintile 4 Quintile 700/136,283 0.90 (0.79–1.04) 0.90 419/82,733 1.00 (0.83–1.20) 0.92 (0.78–1.08) 291/58,702 0.99 (0.55–1.77) 320/78,951 0.97 (0.82–1.15) 0.97 28/5,431 276/53,775 0.92 (0.64–1.33) 0.92 0.93 (0.83–1.04) 0.93 83/9,635 Quintile 3 Quintile 682/136,005 0.86 (0.75–0.98) 0.86 422/87,899 0.99 (0.83–1.19) 0.99 0.92 (0.79–1.07) 0.92 256/53,324 1.07 (0.61–1.89) 1.07 325/79,132 0.87 (0.74–1.03) 0.87 28/5,541 1.00 (0.70–1.43) 0.90 (0.81–1.00) 0.90 77/8,315 Quintile 2 Quintile 650/135,682 1 533/94,800 1 1 230/45,932 1 383/80,145 1 26/4,410 326/58,127 326/54,737 1 1 54/5,850 Quintile 1 Quintile 737/136,322 years years years years years years years HR (95% CI) Cases/p- HR (95% CI) HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- Cases/p- HR (95% CI) HR (95% CI) Cases/p- Cases/p- Yes Prudent diet BMI (normal)18.5–24.9 Dentist No Never Yes Former Smoking at baseline at Smoking Current Subgroup TABLE 3 (Continued)TABLE ALHASSANI et al. | 9

‡ (Continues) .34 .29 (Obese vs Non-obese) vs (Obese .91 interaction p

† .48 .99 .34 .86 .34 .60 p Trend .81 .27 1.03 (0.92–1.16) 1.03 745/136,720 0.96 (0.81–1.14) 0.96 450/88,361 1.08 (0.92–1.26) 332/53,690 1.06 (0.88–1.28) 373/61,620 1.04 (0.89–1.20) 409/80,430 0.87 (0.64–1.18) 0.87 103/15,622 1.09 (0.92–1.29) 1.09 1.03 (0.85–1.23) Quintile 5 Quintile 333/64,802 343/60,986 1.08 (0.96–1.20) 743/136,090 1.00 (0.84–1.17) 407/80,813 1.10 (0.95–1.27) 1.10 368/61,106 1.05 (0.87–1.25) 1.05 340/59,633 1.10 (0.96–1.27) 1.10 435/82,286 0.83 (0.61–1.12) 0.83 94/15,228 1.14 (0.97–1.34) 1.14 1.05 (0.88–1.25) Quintile 4 Quintile 318/61,972 362/64,214 0.99 (0.88–1.10) 676/136,127 0.92 (0.78–1.08) 333/73,240 1.01 (0.88–1.17) 372/68,454 1.01 (0.84–1.21) 1.01 300/55,578 0.98 (0.86–1.12) 0.98 405/86,115 0.80 (0.59–1.08) 85/15,618 1.06 (0.91–1.24) 1.06 0.94 (0.79–1.12) 0.94 Quintile 3 Quintile 269/58,694 350/66,908 1.03 (0.93–1.15) 0.95 (0.80–1.12) 0.95 312/65,909 1.07 (0.93–1.22) 1.07 426/75,247 1.12 (0.94–1.34) 1.12 301/50,556 0.98 (0.86–1.12) 0.98 437/90,599 0.79 (0.59–1.05) 0.79 1.05 (0.91–1.22) 1.05 1.06 (0.89–1.26) 1.06 Quintile 2 Quintile 286/55,385 357/68,322 1 707/135,536 707/135,894 1 271/54,655 1 467/85,699 1 224/41,253 1 514/99,102 1 109/17,337 92/17,014 1 1 Quintile 1 Quintile 270/52,628 357/69,856 years years years years years years years years HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) Cases/p- HR (95% CI) HR (95% CI) Cases/p- Cases/p- Diabetes No ≥Median Physical activityPhysical

‡ .09 .18 .88 interaction p

† .73 .30 .04 .75 .87 p Trend .91 0.98 (0.84–1.14) 0.98 1.06 (0.89–1.26) 1.06 486/91,211 390/81,731 1.17 (0.98–1.39) 1.17 1.04 (0.87–1.25) 296/50,839 324/54,248 0.99 (0.56–1.75) 0.99 0.93 (0.68–1.28) 68/6,072 Quintile 5 Quintile 37/5,330 1.03 (0.89–1.19) 1.03 1.06 (0.90–1.25) 1.06 469/85,990 362/79,701 1.14 (0.96–1.34) 1.14 1.07 (0.90–1.27) 1.07 306/55,928 326/55,244 0.83 (0.48–1.42) 1.08 (0.82–1.43) 87/6,974 Quintile 4 Quintile 32/5,829 1.00 (0.87–1.16) 0.89 (0.75–1.05) 0.89 429/81,758 296/79,303 0.95 (0.81–1.12) 0.95 1.07 (0.91–1.27) 1.07 276/59,935 316/53,351 0.82 (0.48–1.41) 0.97 (0.74–1.27) 0.97 93/9,040 Quintile 3 Quintile 29/5,566 1.01 (0.88–1.17) 1.01 (0.86–1.18) 1.01 411/76,155 321/77,342 1.03 (0.89–1.20) 1.03 1.05 (0.89–1.24) 1.05 327/65,001 297/51,864 0.85 (0.50–1.44) 0.85 0.98 (0.77–1.25) 0.98 120/11,950 Quintile 2 Quintile 31/5,262 1 1 375/68,497 292/71,092 1 1 363/71,858 275/51,275 1 1 171/17,986 Quintile 1 Quintile 31/4,818 years years years years years years HR (95% CI) HR (95% CI) Cases/p- Cases/p- HR (95% CI) HR (95% CI) Cases/p- Cases/p- HR (95% CI) HR (95% CI) Cases/p- Cases/p- -value for the interaction term between an indicator variable for the stratifying term and the continuous variable for dietary pattern (that was used for test of trend). -value when each quintile was assigned the median value and treated as a continuous variable. Yes Never Dentist No Former Smoking at baseline at Smoking Current Subgroup Yes p p Note: Models adjusted for age, energy intake (quintiles), smoking (CSI), physical activity 30+), occupation (METs quintiles), 15–29, (dentist alcohol vs 5–14.9, non-dentist), (g/day: 0, 0.1–4.9, Black/Asian/Other) 30+), except race for (White/ the stratification 25–29.9, and BMI 18.5–24.9, (<18.5, variable. † ‡ TABLE 3 (Continued)TABLE ALHASSANI et al. | 11

1.35* has been well documented (Chaffee & Weston, 2010; Nascimento 1.13 1.12 1.05 et al., 2015; Suvan et al., 2011). Causality, however, is not estab- 1.5 1.02 0.85 1.01 1.02 0.91 0.95 lished (Shungin et al., 2015). Obesity is reported to be associated 1.00 1.00 0.85* 0.89 0.89 1 with “metainflammation” which is defined as a state of low-grade chronic inflammation orchestrated by cells in response to exces- 0.5 Obese sive intake of nutrients and energy (Gregor & Hotamisligil, 2011). Overweight Normal weight A Western diet is positively associated with systemic levels of in- 0 Q1 Q2 Q3 Q4 Q5 flammatory biomarkers (Barbaresko et al., 2013; Esmaillzadeh et al., 2007a; Lopez-Garcia et al., 2004). We believe that the association Normal weight Overweight Obese observed among obese in our study is strongly germane to modification of the host response through systemic inflammation. Pink et al. FIGURE 1 Multivariable HRs for periodontal disease by joint classification of Western diet and BMI categories. Adjusted for found a positive association between levels of fibrinogen and white age, energy intake (quintiles), smoking (CSI), physical activity (METs blood cell (WBC), as markers for systemic inflammation, at baseline quintiles), alcohol (g/day: 0, 0.1–4.9, 5–14.9, 15–29, 30+), occupation and the degree of breakdown of periodontal tissue during the 11- (dentist vs non-dentist) and race (White/Black/Asian/Other). year follow-up of the study (Pink et al., 2015). A Western diet seems Participants in this analysis were classified into 15 groups according to exacerbate the low-grade inflammatory status in obese individ- to the combination of their BMI and their diet score; the reference uals; this increase in inflammatory milieu, due to the concomitant group were those with a combination of the lowest BMI (18.5–24.9) and the lowest Western diet quintile. *p-value <.05 effects of obesity and diet, could strongly alter the host response, and hence the risk of periodontitis. components such as the stage and grade of destruction (Caton et al., Another explanation of the effect modification by obesity is the 2018). It is empirically possible that diet in our study could have an fact that obesity and weight gain are strongly associated with a higher impact that our self-reported periodontal measure did not detect (e.g. risk of diabetes (Koh-Banerjee et al., 2004). While we controlled for impact on differences in probing depth, clinical attachment loss and/ diabetes by several methods in the sensitivity analysis, one cannot or clinical inflammation). exclude the possibility that unreported pre-diabetes insulin resis- Among obese individuals, we observed a higher incidence of tance and glucose intolerance may have mediated the impact of diet periodontitis in the highest compared to the lowest quintiles of on periodontal health, as Western diet is positively associated with Western diet. Given that adiposity is a predictor of periodontal insulin resistance even after controlling for BMI (Esmaillzadeh et al., disease (Jimenez et al., 2012a) and that Western dietary pattern 2007b). Underlying insulin resistance tends to intensify the adverse is associated with weight gain (Schulze et al., 2006), BMI could po- metabolic effects of diet, potentially through the increased insulin tentially mediate, confound, and/or act as an effect modifier in the secretion needed to metabolize ingested carbohydrates (Jeppesen relationship between dietary patterns and periodontitis. Since the et al., 1997; Liu et al., 2000; Miller, 1994; Tabung et al., 2016). Also, association was limited to obese adults, and since adjusting for BMI insulin resistance and impaired glucose metabolism have been as- did not change the results of our analysis, obesity could be an effect sociated with periodontal disease (Andriankaja & Joshipura, 2014; modifier, but BMI does not seem to be a confounder or a mediator Benguigui et al., 2010; Islam et al., 2015; Timonen et al., 2013). in these analyses. This study is a secondary analysis of an ongoing cohort estab- Diet may affect periodontal health through a number of mech- lished for other health outcomes; hence, we made additional exclu- anisms (Boyd & Madden, 2003). Consumption of highly processed sions specific to our research question. Men with periodontitis and and/or sugary food items is associated with postprandial systemic edentulous men at baseline who were not at risk of developing the inflammation and oxidative stress, which may create an environment outcome were not included. We excluded participants with major conducive to periodontal breakdown (Chapple, 2009). Poor quality health issues at baseline as their reported diet may not be repre- diet is also associated with impaired glucose tolerance which is a risk sentative of their long-term diet. These exclusions were made to factor for periodontal disease (Genco et al., 2005). Another mecha- improve internal validity of our study, but may limit generalizabil- nism is via the obesogenic effect of food that may lead to increased ity. The main strength of our project is the large cohort with long body fatness, hence increased periodontal disease risk (Chapple, follow-up, validated methods for collecting dietary data and the 2009). Additionally, diet may directly influence oral bacterial activ- fact that participants are highly educated and motivated, which ity and the proportions of different species during bacterial biofilm minimizes the risk of information bias. To our knowledge, this is the development (Bowden & Li, 1997). Less established mechanisms in- largest study, with the longest follow-up evaluating the relationship volve the potential effect of diet on modifying intestinal microbiome between diet and periodontal disease. Our study, however, has sev- into a more Gram-negative bacterial flora, which is associated with eral limitations. Due to the observational nature of the study, un- changes in the oral microbiota in animal models (Burcelin et al., 2011). measured and residual confounding cannot be ruled out. In addition, The observed association between Western diet and periodontal owing to the homogeneity of the study population (mainly Caucasian disease limited to obese in our study warrants further investigation. male health professionals), the generalizability of the results may be The relationship of increased body fatness with higher periodontitis limited. On the other hand, the homogeneity of the cohort can be 12 | ALHASSANI et al. viewed as a strength, as confounding by socio-economic factors was Health Nutrition, 19, 503-510. https://doi.org/10.1017/s136898001 5001202 minimized, thus improving the internal validity of the study. Another Adegboye, A. R., Christensen, L. B., Holm-Pedersen, P., Avlund, K., limitation is the self-report method for data collection, which due to Boucher, B. J., & Heitmann, B. L. (2012). Intake of dairy products in their inherent measurement error could have attenuated the results. relation to periodontitis in older Danish adults. Nutrients, 4, 1219- However, the validity of self-administered FFQs and self-reported 1229. https://doi.org/10.3390/nu4091219 Albandar, J. M. (2005). Epidemiology and risk factors of periodontal dis- periodontal disease has been evaluated previously, and both meth- eases. Dental Clinics of North America, 49(3), 517–532. https://doi. ods were proven as acceptable approaches. In addition, we used the org/10.1016/j.cden.2005.03.003 cumulative averages of each dietary pattern score as the main ex- Al-Zahrani, M. S. (2006). Increased intake of dairy products is related to posure, which in addition to the energy adjustment in our analysis, lower periodontitis prevalence. Journal of Periodontology, 77, 289- should mitigate measurement error. 294. https://doi.org/10.1902/jop.2006.050082 Al-Zahrani, M. S., Bissada, N. F., & Borawskit, E. A. (2003). Obesity In conclusion, we observed no overall association be- and periodontal disease in young, middle-aged, and older adults. tween prudent or Western diet and risk of periodontal disease. Journal of Periodontology, 74, 610-615. https://doi.org/10.1902/ Subgroups analysis showed that among obese men, those with jop.2003.74.5.610 high scores of Western dietary pattern had a significantly higher Al-Zahrani, M. S., Borawski, E. A., & Bissada, N. F. (2004). Poor overall diet quality as a possible contributor to calculus formation. Oral risk of periodontal disease compared to those with lower scores. Health and Preventive Dentistry, 2, 345-349. Our findings support the hypothesis that diet is a modifiable risk Al-Zahrani, M. S., Borawski, E. A., & Bissada, N. F. (2005a). factor of periodontal disease, at least among obese individuals. Increased physical activity reduces prevalence of periodonti- Replication studies are warranted, and future research may focus tis. Journal of Dentistry, 33, 703-710. https://doi.org/10.1016/j. jdent.2005.01.004 on the mechanisms through which diet may influence periodontal Al-Zahrani, M. S., Borawski, E. A., & Bissada, N. F. (2005b). Periodontitis health. Also, it would be of value to explore the components of and three health-enhancing behaviors: maintaining normal weight, diet most relevant to periodontal health, and whether their poten- engaging in recommended level of exercise, and consuming a tial effect is generalizable to other subgroups in the population, high-quality diet. Journal of Periodontology, 76, 1362-1366. https:// doi.org/10.1902/jop.2005.76.8.1362 besides obese. Andriankaja, O. M., & Joshipura, K. (2014). Potential association between prediabetic conditions and gingival and/or periodontal inflam- ACKNOWLEDGMENTS mation. Journal of Diabetes Investigation, 5, 108-114. https://doi. The authors thank the participants and staff of the Health org/10.1111/jdi.12122 Ardisson Korat, A. V., Willett, W. C., & Hu, F. B. (2014). Diet, lifestyle, and Professionals Follow-Up Study. This article has been supported by genetic risk factors for type 2 diabetes: A review from the Nurses' research grant UM1 CA167552, Health Professionals Follow-Up Health Study, Nurses' Health Study 2, and Health Professionals' Study infrastructure grant, from the National Institutes of Health Follow-up Study. Current Nutrition Reports, 3, 345-354. https://doi. (NIH). org/10.1007/s13668-014-0103-5 Astrup, A., Dyerberg, J., Elwood, P., Hermansen, K., Hu, F. B., Jakobsen, M. U., Kok, F. J., Krauss, R. M., Lecerf, J. M., LeGrand, CONFLICT OF INTEREST P., Nestel, P., Riserus, U., Sanders, T., Sinclair, A., Stender, S., The authors had no conflict of interest to disclose. Tholstrup, T., & Willett, W. C. (2011). The role of reducing intakes of saturated fat in the prevention of cardiovascular dis- ETHICS STATEMENT ease: where does the evidence stand in 2010? American Journal of Clinical Nutrition, 93, 684-688. https://doi.org/10.3945/ The study was approved by the institutional review boards of the ajcn.110.004622 Brigham and Women's Hospital and the Harvard T. H. Chan School Barbaresko, J., Koch, M., Schulze, M. B., & Nothlings, U. (2013). Dietary of Public Health. Participants' completion of the questionnaires con- pattern analysis and biomarkers of low-grade inflammation: a sys- stituted informed consent. tematic literature review. Nutrition Reviews, 71, 511-527. https:// doi.org/10.1111/nure.12035 Benguigui, C., Bongard, V., Ruidavets, J. B., Chamontin, B., Sixou, M., DATA AVAILABILITY STATEMENT Ferrieres, J., & Amar, J. (2010). Metabolic syndrome, insulin resis- Further information including the procedures to obtain and access tance, and periodontitis: a cross-sectional study in a middle-aged data from the Health Professionals Follow-up Study is described at French population. Journal of Clinical Periodontology, 37, 601-608. https://doi.org/10.1111/j.1600-051X.2010.01571.x https://sites.sph.harvard.edu/hpfs/for-collaborators/. Blicher, B., Joshipura, K., & Eke, P. (2005). Validation of self-reported periodontal disease: A systematic review. Journal of Dental ORCID Research, 84, 881-890. https://doi.org/10.1177/1544059105 Ahmed A. Alhassani https://orcid.org/0000-0002-3007-0128 08401003 Bowden, G. H., & Li, Y. H. (1997). Nutritional influences on biofilm de- Kaumudi J. Joshipura https://orcid.org/0000-0003-1964-7579 velopment. Advances in Dental Research, 11, 81-99. https://doi. org/10.1177/08959374970110012101 REFERENCES Boyd, L. D., & Madden, T. E. (2003). Nutrition, infection, and periodontal Adegboye, A. R., Boucher, B. J., Kongstad, J., Fiehn, N. E., Christensen, disease. Dental Clinics of North America, 47, 337-354. L. B., & Heitmann, B. L. (2016). Calcium, vitamin D, casein and Burcelin, R., Serino, M., Chabo, C., Blasco-Baque, V., & Amar, J. whey protein intakes and periodontitis among Danish adults. Public (2011). Gut microbiota and diabetes: from pathogenesis to ALHASSANI et al. | 13

therapeutic perspective. Acta Diabetologica, 48, 257-273. https:// Iwasaki, M., Yoshihara, A., Moynihan, P., Watanabe, R., Taylor, G. W., doi.org/10.1007/s00592-011-0333-6 & Miyazaki, H. (2010). Longitudinal relationship between dietary Caton, J. G., Armitage, G., Berglundh, T., Chapple, I. L. C., Jepsen, omega-3 fatty acids and periodontal disease. Nutrition, 26, 1105- S., Kornman, K. S., Mealey, B. L., Papapanou, P. N., Sanz, M., & 1109. https://doi.org/10.1016/j.nut.2009.09.010 Tonetti, M. S. (2018). A new classification scheme for periodon- Jeppesen, J., Schaaf, P., Jones, C., Zhou, M. Y., Chen, Y. D., & Reaven, tal and peri-implant diseases and conditions – Introduction and G. M. (1997). Effects of low-fat, high-carbohydrate diets on risk key changes from the 1999 classification. Journal of Clinical factors for ischemic heart disease in postmenopausal women. Periodontology, 45(Suppl 20), S1-S8. https://doi.org/10.1111/ American Journal of Clinical Nutrition, 65, 1027-1033. https://doi. jcpe.12935 org/10.1093/ajcn/65.4.1027 Chaffee, B. W., & Weston, S. J. (2010). Association between chronic Jimenez, M., Giovannucci, E., Krall Kaye, E., Joshipura, K. J., & Dietrich, periodontal disease and obesity: A systematic review and me- T. (2014). Predicted vitamin D status and incidence of tooth loss ta-analysis. Journal of Periodontology, 81, 1708-1724. https://doi. and periodontitis. Public Health Nutrition, 17, 844-852. https://doi. org/10.1902/jop.2010.100321 org/10.1017/s1368980013000177 Chapple, I. L. (2009). Potential mechanisms underpinning the nutritional Jimenez, M., Hu, F. B., Marino, M., Li, Y., & Joshipura, K. J. (2012a). modulation of periodontal inflammation. Journal of the American Prospective associations between measures of adiposity and peri- Dental Association, 140, 178-184. odontal disease. Obesity (Silver Spring), 20, 1718-1725. https://doi. Dietrich, T., & Hoffmann, K. (2004). A comprehensive index for the mod- org/10.1038/oby.2011.291 eling of smoking history in periodontal research. Journal of Dental Jimenez, M., Hu, F. B., Marino, M., Li, Y., & Joshipura, K. J. (2012b). Type Research, 83, 859-863. https://doi.org/10.1177/1544059104 2 diabetes mellitus and 20 year incidence of periodontitis and tooth 08301107 loss. Diabetes Research and Clinical Practice, 98, 494-500. https:// Esmaillzadeh, A., Kimiagar, M., Mehrabi, Y., Azadbakht, L., Hu, F. B., & doi.org/10.1016/j.diabres.2012.09.039 Willett, W. C. (2007a). Dietary patterns and markers of systemic Joshipura, K. J., & Andriankaja, O. M. (2016). Modifiable systemic factors inflammation among Iranian women. Journal of Nutrition, 137, 992- for periodontal disease prevention and management. In E. Wallace 998. https://doi.org/10.1093/jn/137.4.992 (Ed.), Periodontal disease: Diagnosis, management options and clinical Esmaillzadeh, A., Kimiagar, M., Mehrabi, Y., Azadbakht, L., Hu, F. B., features (pp. 47–85). New York: Nova Science Publishers. & Willett, W. C. (2007b). Dietary patterns, insulin resistance, Joshipura, K. J., Douglass, C. W., Garcia, R. I., Valachovic, R., & Willett, W. and prevalence of the metabolic syndrome in women. American C. (1996). Validity of a self-reported periodontal disease measure. Journal of Clinical Nutrition, 85, 910-918. https://doi.org/10.1093/ Journal of Public Health Dentistry, 56, 205-212. ajcn/85.3.910 Joshipura, K. J., Pitiphat, W., & Douglass, C. W. (2002). Validation of Fujita, Y., & Maki, K. (2015). High-fat diet-induced obesity triggers alveo- self-reported periodontal measures among health professionals. lar bone loss and spontaneous periodontal disease in growing mice. Journal of Public Health Dentistry, 62, 115-121. BMC Obesity, 3, 1. https://doi.org/10.1186/s40608-016-0082-8 Kaye, E. K. (2012). Nutrition, dietary guidelines and optimal peri- Fung, T. T., Schulze, M., Manson, J. E., Willett, W. C., & Hu, F. B. (2004). odontal health. Periodontology 2000, 58, 93-111. https://doi. Dietary patterns, meat intake, and the risk of type 2 diabetes in org/10.1111/j.1600-0757.2011.00418.x women. Archives of Internal Medicine, 164, 2235-2240. https://doi. Kiefte-de Jong, J. C., Li, Y., Chen, M., Curhan, G. C., Mattei, J., Malik, V. S., org/10.1001/archinte.164.20.2235 Forman, J. P., Franco, O. H., & Hu, F. B. (2017). Diet-dependent acid Genco, R. J., Grossi, S. G., Ho, A., Nishimura, F., & Murayama, Y. (2005). A load and type 2 diabetes: Pooled results from three prospective proposed model linking inflammation to obesity, diabetes, and peri- cohort studies. Diabetologia, 60, 270-279. https://doi.org/10.1007/ odontal infections. Journal of Periodontology, 76(Suppl 11S), 2075- s00125-016-4153-7 2084. https://doi.org/10.1902/jop.2005.76.11-S.2075 Koh-Banerjee, P., Wang, Y., Hu, F. B., Spiegelman, D., Willett, W. C., & Gregor, M. F., & Hotamisligil, G. S. (2011). Inflammatory mechanisms Rimm, E. B. (2004). Changes in body weight and body fat distri- in obesity. Annual Review of Immunology, 29, 415-445. https://doi. bution as risk factors for clinical diabetes in US men. American org/10.1146/annurev-immun ol-031210-101322 Journal of Epidemiology, 159, 1150-1159. https://doi.org/10.1093/ Heidemann, C., Schulze, M. B., Franco, O. H., van Dam, R. M., Mantzoros, aje/kwh167 C. S., & Hu, F. B. (2008). Dietary patterns and risk of mortality from Leffondre, K., Abrahamowicz, M., Siemiatycki, J., & Rachet, B. (2002). cardiovascular disease, cancer, and all causes in a prospective co- Modeling smoking history: A comparison of different approaches. hort of women. Circulation, 118, 230-237. https://doi.org/10.1161/ American Journal of Epidemiology, 156, 813-823. circulationaha.108.771881 Liu, S., Willett, W. C., Stampfer, M. J., Hu, F. B., Franz, M., Sampson, L., Holtfreter, B., Albandar, J. M., Dietrich, T., Dye, B. A., Eaton, K. A., Eke, Hennekens, C. H., & Manson, J. E. (2000). A prospective study of P. I., Papapanou, P. N., & Kocher, T. (2015). Standards for report- dietary glycemic load, carbohydrate intake, and risk of coronary ing chronic periodontitis prevalence and severity in epidemiologic heart disease in US women. American Journal of Clinical Nutrition, studies: Proposed standards from the Joint EU/USA Periodontal 71, 1455-1461. https://doi.org/10.1093/ajcn/71.6.1455 Epidemiology Working Group. Journal of Clinical Periodontology, 42, Lopez-Garcia, E., Schulze, M. B., Fung, T. T., Meigs, J. B., Rifai, N., Manson, 407-412. https://doi.org/10.1111/jcpe.12392 J. E., & Hu, F. B. (2004). Major dietary patterns are related to Hu, F. B. (2002). Dietary pattern analysis: A new direction in nutritional plasma concentrations of markers of inflammation and endothelial epidemiology. Current Opinion in Lipidology, 13, 3-9. dysfunction. American Journal of Clinical Nutrition, 80, 1029-1035. Hu, F. B., Rimm, E., Smith-Warner, S. A., Feskanich, D., Stampfer, M. J., https://doi.org/10.1093/ajcn/80.4.1029 Ascherio, A., Sampson, L., & Willett, W. C. (1999). Reproducibility Malik, V. S., Fung, T. T., van Dam, R. M., Rimm, E. B., Rosner, B., & Hu, F. and validity of dietary patterns assessed with a food-frequency B. (2012). Dietary patterns during adolescence and risk of type 2 questionnaire. American Journal of Clinical Nutrition, 69, 243-249. diabetes in middle-aged women. Diabetes Care, 35, 12-18. https:// https://doi.org/10.1093/ajcn/69.2.243 doi.org/10.2337/dc11-0386 Islam, S. K., Seo, M., Lee, Y. S., & Moon, S. S. (2015). Association of peri- Merchant, A. T., Pitiphat, W., Franz, M., & Joshipura, K. J. (2006). odontitis with insulin resistance, beta-cell function, and impaired Whole-grain and fiber intakes and periodontitis risk in men. fasting glucose before onset of diabetes. Endocrine Journal, 62, 981- American Journal of Clinical Nutrition, 83, 1395-1400. https://doi. 989. https://doi.org/10.1507/endocrj.EJ15-0350 org/10.1093/ajcn/83.6.1395 14 | ALHASSANI et al.

Meyerhardt, J. A., Niedzwiecki, D., Hollis, D., Saltz, L. B., Hu, F. B., Mayer, Shimazaki, Y., Shirota, T., Uchida, K., Yonemoto, K., Kiyohara, Y., Iida, M., R. J., Nelson, H., Whittom, R., Hantel, A., Thomas, J., & Fuchs, C. S. Saito, T., & Yamashita, Y. (2008). Intake of dairy products and peri- (2007). Association of dietary patterns with cancer recurrence and odontal disease: the Hisayama Study. Journal of Periodontology, 79, survival in patients with stage III colon cancer. JAMA, 298, 754-764. 131-137. https://doi.org/10.1902/jop.2008.070202 https://doi.org/10.1001/jama.298.7.754 Shungin, D., Cornelis, M. C., Divaris, K., Holtfreter, B., Shaffer, J. R., Miller, J. C. (1994). Importance of glycemic index in diabetes. American Yu, Y. H., Barros, S. P., Beck, J. D., Biffar, R., Boerwinkle, E. A., Journal of Clinical Nutrition, 59, 747s-752s. https://doi.org/10.1093/ Crout, R. J., Ganna, A., Hallmans, G., Hindy, G., Hu, F. B., Kraft, P., ajcn/59.3.747S McNeil, D. W., Melander, O., Moss, K. L., … Franks, P. W. (2015). Muluke, M., Gold, T., Kiefhaber, K., Al-Sahli, A., Celenti, R., Jiang, H., Using genetics to test the causal relationship of total adiposity Cremers, S., Van Dyke, T., & Schulze-Spate, U. (2016). Diet-induced and periodontitis: Mendelian randomization analyses in the Gene- obesity and its differential impact on periodontal bone loss. Journal Lifestyle Interactions and Dental Endpoints (GLIDE) Consortium. of Dental Research, 95, 223-229. https://doi.org/10.1177/00220 International Journal of Epidemiology, 44, 638-650. https://doi. 34515609882 org/10.1093/ije/dyv075 Nascimento, G. G., Leite, F. R., Do, L. G., Peres, K. G., Correa, M. B., Suvan, J., D'Aiuto, F., Moles, D. R., Petrie, A., & Donos, N. (2011). Demarco, F. F., & Peres, M. A. (2015). Is weight gain associated Association between overweight/obesity and periodontitis in with the incidence of periodontitis? A systematic review and me- adults. A systematic review. Obesity Reviews, 12, e381-e404. ta-analysis. Journal of Clinical Periodontology, 42, 495-505. https:// https://doi.org/10.1111/j.1467-789X.2010.00808.x doi.org/10.1111/jcpe.12417 Tabung, F. K., Wang, W., Fung, T. T., Hu, F. B., Smith-Warner, S. A., Nielsen, S. J., Trak-Fellermeier, M. A., Joshipura, K., & Dye, B. A. (2016). Chavarro, J. E., Fuchs, C. S., Willett, W. C., & Giovannucci, E. L. Dietary fiber intake is inversely associated with periodontal disease (2016). Development and validation of empirical indices to assess among US adults. Journal of Nutrition, 146, 2530-2536. https://doi. the insulinaemic potential of diet and lifestyle. British Journal of org/10.3945/jn.116.237065 Nutrition, 116(10), 1787–1798. https://doi.org/10.1017/s0007 Otomo-Corgel, J., Pucher, J. J., Rethman, M. P., & Reynolds, M. A. (2012). 114516003755 State of the science: chronic periodontitis and systemic health. Timonen, P., Saxlin, T., Knuuttila, M., Suominen, A. L., Jula, A., Tervonen, Journal of Evidence Based Dental Practice, 12, 20-28. https://doi. T., & Ylostalo, P. (2013). Role of insulin sensitivity and beta cell func- org/10.1016/s1532-3382(12)70006-4 tion in the development of periodontal disease in adults without Pink, C., Kocher, T., Meisel, P., Dorr, M., Markus, M. R., Jablonowski, L., diabetes. Journal of Clinical Periodontology, 40, 1079-1086. https:// Grotevendt, A., Nauck, M., & Holtfreter, B. (2015). Longitudinal doi.org/10.1111/jcpe.12162 effects of systemic inflammation markers on periodontitis. Journal van Dam, R. M., Rimm, E. B., Willett, W. C., Stampfer, M. J., & Hu, F. of Clinical Periodontology, 42, 988-997. https://doi.org/10.1111/ B. (2002). Dietary patterns and risk for type 2 diabetes mellitus jcpe.12473 in U.S. men. Annals of Internal Medicine, 136, 201-209. https://doi. Pitiphat, W., Merchant, A. T., Rimm, E. B., & Joshipura, K. J. (2003). org/10.7326/0003-4819-136-3-200202050-00008 Alcohol consumption increases periodontitis risk. Journal of Dental Willett, W. (2012). Nutritional epidemiology (pp. ix). Oxford University Research, 82, 509-513. https://doi.org/10.1177/1544059103 Press. 08200704 Yoshihara, A., Watanabe, R., Hanada, N., & Miyazaki, H. (2009). A longitu- Qi, L., Cornelis, M. C., Zhang, C., van Dam, R. M., & Hu, F. B. (2009). dinal study of the relationship between diet intake and dental caries Genetic predisposition, Western dietary pattern, and the risk of and periodontal disease in elderly Japanese subjects. Gerodontology, type 2 diabetes in men. American Journal of Clinical Nutrition, 89, 26, 130-136. https://doi.org/10.1111/j.1741-2358.2008.00244.x 1453-1458. https://doi.org/10.3945/ajcn.2008.27249 Rimm, E. B., Giovannucci, E. L., Stampfer, M. J., Colditz, G. A., Litin, L. B., & Willett, W. C. (1992). Reproducibility and validity of an expanded SUPPORTING INFORMATION self-administered semiquantitative food frequency questionnaire Additional supporting information may be found online in the among male health professionals. American Journal of Epidemiology, Supporting Information section. 135, 1114-1126. discussion 1127–1136. Rothman, K. J., Greenland, S., & Lash, T. L. (2008). Validity in epidemiologic studies. In K. J. Rothman, S. Greenland, & T. L. Lash (Eds.), Modern epidemiology (3rd ed., pp. 128-147). Wolters Kluwer Health/ Lippincott Williams & Wilkins. How to cite this article: Alhassani AA, Hu FB, Li Y, Rosner Schulze, M. B., Fung, T. T., Manson, J. E., Willett, W. C., & Hu, F. B. BA, Willett WC, Joshipura KJ. The associations between (2006). Dietary patterns and changes in body weight in women. major dietary patterns and risk of periodontitis. J Clin Obesity (Silver Spring), 14, 1444-1453. https://doi.org/10.1038/ oby.2006.164 Periodontol. 2021;48:2–14. https://doi.org/10.1111/ Schwartz, N., Kaye, E. K., Nunn, M. E., Spiro, A. 3rd, & Garcia, R. I. (2012). jcpe.13380 High-fiber foods reduce periodontal disease progression in men aged 65 and older: The Veterans Affairs normative aging study/ Dental Longitudinal Study. Journal of the American Geriatrics Society, 60, 676-683. https://doi.org/10.1111/j.1532-5415.2011.03866.x Received: 10 June 2020 | Revised: 18 August 2020 | Accepted: 24 September 2020 DOI: 10.1111/jcpe.13378

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

Obesity as a risk factor for tooth loss over 5 years: A population-based cohort study

1Periodontology, Faculty of Dentistry, Federal University of Rio Grande do Sul, Abstract Porto Alegre, Brazil Aim: To assess obesity as a risk factor for tooth loss over 5 years in an urban sample 2Periodontal Diagnostics Research Laboratory, Department of Periodontology of Brazilian adults. and Oral Implantology, Temple University Materials and methods: A total of 1586 individuals were surveyed using a multistage School of Dentistry, Philadelphia, PA, USA probabilistic approach. Five years later, 635 individuals 14-64 years old were re-ex- 3Department of Periodontology, Adams School of Dentistry, University of North amined. An incident case of tooth loss was determined for a participant that had lost Carolina at Chapel Hill, Chapel Hill, NC, at least one tooth over time. Obesity was evaluated by calculating body mass index at USA baseline and by the change in obesity status over time. Correspondence Results: Incident cases of tooth loss were significantly more frequent among obese Alex Nogueira Haas, Rua Ramiro Barcelos 2492, Porto Alegre, Brazil, 91035-003. (47.1%) than normal-weight individuals (32.4%) (p = .004). Obese individuals had 31% Email: [email protected] higher risk [relative risk (RR) =1.31; 95% confidence interval (95%CI) 1.04-1.65] for

Funding information tooth loss than normal-weight individuals adjusting for age, socio-economic status, This study was funded by Coordenação smoking, dental care and periodontitis. This association was significant for females de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil (grant #1614/99- (RR=1.47, 95%CI 1.08-2.01), but not for males. The risk for tooth loss was also modi- 1) and Foundation for Research Support fied by presence of periodontitis at baseline and lifetime smoking exposure. There of Rio Grande do Sul (FAPERGS), Porto Alegre, Brazil (grant #PPSUS-0700481). was an increased risk for tooth loss for those that remained obese than those that remained normal weight. Conclusion: Obesity is associated with higher risk for tooth loss. This association was modified by sex, periodontal status and smoking.

KEYWORDS cohort, obesity, periodontitis, risk factor, smoking, tooth loss

1 | INTRODUCTION over time in various countries (Dye et al. 2019; Muller et al. 2007). Caries and periodontitis are recognized as the main causes for tooth Tooth loss has a negative impact on individuals' life and well-being loss, but other factors have also been associated with its occurrence (Emami et al. 2013). Globally, total tooth loss is estimated to affect (Haworth et al. 2018; Seerig et al. 2015; Susin et al. 2005). 267 million individuals and account for 7.3 million years lived with Obesity may be defined as abnormal or excessive fat accumula- disability (Collaborators et al. 2020). Although estimates of severe tion that presents risk to health (Haslam & James 2005). The global tooth loss generally have decreased (Kassebaum et al. 2014), pop- prevalence of obesity is approximately 20% (Bluher 2019). Obesity ulation-based studies still demonstrate high number of lost teeth leads to increased systemic inflammation mediated by the adipose

wileyonlinelibrary.com/journal/jcpe | 15 J Clin Periodontol. 2021;48:15–24. 16 | VALLIM et al. tissue (Ellulu et al. 2017) and is considered a major risk factor for various chronic diseases (G. M. Singh et al. 2013). Obesity has Clinical Relevance also been directly associated with tooth loss. However, the great Scientific rationale for the study: Obesity is a risk factor for majority of studies on this association are cross-sectional (Jiang various chronic diseases, and it has been also associated et al. 2013; Jung et al. 2011; Kang et al. 2019; Ojima et al. 2007; with tooth loss. However, there is very limited evidence Pilotto et al. 2014), and some have evaluated self-reported tooth for a temporal association between obesity and tooth loss. loss (Bernardo Cde et al. 2012; Ostberg et al. 2009). A systematic Principal findings: Obese individuals had higher risk for review (Nascimento et al. 2016) included only four cross-sectional tooth loss over 5 years than normal-weight individuals. studies and found that the chances for tooth loss increased by Effect modification was observed for sex, periodontitis 49% in obese compared to normal-weight individuals. More re- and smoking. Individuals that remained obese over time cently, longitudinal associations between obesity and incident had higher risk for tooth loss than those that remained tooth loss were found in a cohort study in Germany (Meisel et al. normal weight. 2014). Noteworthy, the impact of changes in obese status has not Practical implications: Preventive strategies aiming to re- been addressed so far. duce tooth loss should target obese individuals. The association between obesity and tooth loss may rely on the fact that obesity is also related to periodontitis (Gaio et al. 2016; Nascimento et al. 2015) and probably to dental caries (Vazquez- Nava et al. 2010). Also, the association between obesity and tooth dental unit, including dental chair, light and a compressor, was used loss may be modified by sex (Meisel et al. 2014; A. Singh et al. 2015). in the data collection at both timepoints. Other modifying factors may also be present in this association, but Dental caries and periodontal status were assessed at baseline and have not been evaluated. follow-up for all permanent teeth, except third molars. DMFT index The aim of the present study was to assess obesity as a risk fac- was used to assess dental caries and tooth loss. Probing depth (PD) tor for tooth loss over 5 years in an urban sample of Brazilian adults. and gingival recession (GR) were recorded in six sites per tooth using a manual periodontal probe (PCP10-SE, Hu-Friedy Mfg. Co. Inc). Clinical attachment loss (CAL) was calculated as the sum of PD and GR. 2 | MATERIALS AND METHODS At the end of baseline and follow-up clinical examinations, the height of the participants was measured in centimetres, using The present investigation is a 5-year population-based prospective a stadiometer. Also, weight was assessed in kilograms using me- study, part of the Porto Alegre Study started in 2001 (Haas et al. chanical scales. The scales and the stadiometer were certified by 2012; Susin et al. 2004). Baseline and follow-up data were collected the Brazilian Institute of Metrology, Standardization and Industrial between June and December/2001 and between October/2006 Quality (INMETRO). Certification was performed at the beginning of and January/2007, respectively. This study follows the STROBE the study providing that the scales had a 99% precision. guidelines for reporting observational studies. At baseline, a representative sample of more than 3 million habi- tants from 14 major municipalities in the metropolitan area of Porto 2.2 | Ethical considerations Alegre, Brazil, was obtained using a multistage probability sampling method. In brief, 11 primary sampling units (PSU) were randomly This study was approved by the Institutional Review Board of the selected from geographic areas stratified according to income sta- Federal University of Rio Grande do Sul, Porto Alegre, Brazil (pro- tus. In the second stage, 29 sectors were randomly selected within tocol number 51/05). All participants read and signed an informed each PSU. The third stage included selecting households within each consent before participation in the study. The study was conducted sector. Detailed information on sample characteristics and sampling according to the standards of the Declaration of Helsinki for studies methodology was previously reported (Haas et al. 2012; Susin et al. in humans. 2004).

2.3 | Measurement reproducibility 2.1 | Data collection Detailed information about the measurement reproducibility Interviews were conducted by trained dental assistants using a struc- is published elsewhere (Haas et al. 2012; Susin et al. 2004). In tured written questionnaire. A team of four dentists and two dental brief, reproducibility of interviews and clinical examinations were assistants conducted the fieldwork at baseline. In the follow-up ex- evaluated during the fieldwork at baseline and at the follow-up amination, three dentists and three dental assistants conducted the examination. Interviews were repeated in 79 (5%) and 94 (12.5%) study. One dentist (ANH) participated in both examinations. A mo- participants at baseline and at follow-up, respectively. Overall bile examination unit consisting of a trailer equipped with a complete kappa coefficients for categorical data were ≥0.93. Intra-class VALLIM et al. | 17 correlation coefficients for continuous variables such as smok- defined, in which an incident case was determined for a participant ing exposure (number of packs smoked during lifetime) and num- that had lost at least one tooth over the follow-up period. ber of dental visits during the last 5 years were 0.93 and 0.96, respectively. Repeated measurements of periodontal parameters were per- 2.5 | Obesity formed in 57(3.9%) and 45(6.5%) participants at baseline and follow-up, respectively. The intra-examiner reproducibility revealed Obesity was evaluated by calculating the body mass index (BMI) a high agreement of CAL for the gold-standard examiner (CS) with using two analytical approaches. Firstly, obesity at baseline was weighted kappa coefficients (±1 mm) ranging between 0.86 and defined according to the WHO criteria (Kopelman 2000). Four BMI 0.87 for site measurements. Inter-examiner weighted kappa co- categories were defined: underweight (BMI <18.5 kg/m2), nor- efficient (±1 mm) for CAL ranged between 0.64 and 0.71 for site mal weight (BMI 18.5 - 24.9 kg/m2), overweight (BMI 25 - 29.9 kg/ measurements. Intra-class correlation coefficient ranged between m2) and obese (BMI ≥30 kg/m2). For the purposes of the present 0.95 and 0.98 for mean CAL. The reliability of BMI was evaluated study, individuals classified as underweight were excluded from the by duplicated measurements of weight and height in 42 randomly analyses. selected participants. The Kappa value for categorical BMI was Secondly, change in obesity status over the 5 years was deter- 0.88. mined. Individuals were classified as:

(0) Remained normal weight: those with normal weight at base- 2.4 | Outcome line that remained classified as normal after follow-up (1) Lost weight: those obese at baseline that became over- The primary outcome of this study was tooth loss over 5 years. A weight, and those overweight at baseline that became normal tooth that was no longer present at the follow-up examination was weight (no individuals were overweight/obese and became considered lost over time. A binary outcome for tooth loss was normal weight)

FIGURE 1 Flow chart of study sample 18 | VALLIM et al.

TABLE 1 Incidence of tooth loss (percentage of individuals with (2) Gained weight: those that were normal weight and became at least one lost tooth) according to exposure variables overweight or obese, and those that were overweight that be- Whole sample Percentage came obese (n) (SE) p (3) Remained overweight: those that were overweight at baseline Sex and remained overweight Females 367 35.7 (2.5) Ref. (4) Remained obese: those that were obese at baseline and re- Males 268 36.6 (2.9) .82 mained obese. Age 14-29 years 228 29.8 (3.0) Ref. 30-49 years 267 37.8 (2.9) .06 2.6 | Other exposure variables 50-64 years 140 42.9 (4.2) .03 Education Age, sex, education, socio-economic status, tobacco smoking, diabetes, dental care over 5 years, number of teeth present, presence ≥12 years 113 28.7 (2.8) Ref. of periodontitis and dental caries were exposure variables assessed 5-11 years 264 40.1 (3.0) .56 as possible confounders and modifiers in the association between ≤4 years 258 43.3 (4.6) .01 obesity and tooth loss. These data were derived from the baseline Socio-economic status examination, except for dental care over 5 years that was assessed High 209 27.8 (2.9) Ref. at follow-up. Medium 192 39.1 (3.5) .47 Socio-economic status was scored using the standard Brazilian Low 234 42.6 (3.4) .001 economy classification (CCEB). Socio-economic status was cate- Smoking gorized according to tertiles into low (≤12 CCEB points), middle Never-smokers 359 33.1 (2.5) Ref. (13-17 CCEB points) and high socio-economic status (≥18 CCEB Light–moderate 140 30.7 (2.5) .59 points). Education was categorized into three categories: ≤4 years, smokers 5-11 years and ≥12 years of education. Heavy smokers 136 49.2 (4.3) .001 Dental care was classified according to participants’ self-re- Dental care ported frequency of dental visits from baseline (2001) to fol- No 88 25.0 (4.6) Ref. low-up (2006). Individuals who have not visited a dentist during Irregular 263 35.7 (2.9) .05 the last 5 years were classified as having no dental care. Those vis- Regular 284 39.8 (2.9) .007 iting a dentist 1-4 times during the 5-year follow-up period were Diabetes classified as having irregular dental care. Individuals with regular No 604 35.6 (19.5) Ref. dental care were those with at least 5 dental visits during the fol- Yes 31 45.1 (90.8) .30 low-up period. Periodontitis Lifetime exposure to cigarette smoking (pack-years) was calcu- No 444 28.1 (21.3) Ref. lated by multiplying the number of packs of cigarettes consumed per Yes 191 54.4 (36.1) <.001 day by the number of years of habit. Three categories of smoking Dental caries exposure were determined: never-smokers, light–moderate smokers DMF T<=12 288 29.1 (26.8) Ref. (1-14 pack-years) and heavy smokers (≥15 pack-years). DMFT >=13 347 41.8 (26.5) .001 The presence of periodontitis at baseline was defined as pre- Number of teeth at baseline viously reported using thresholds of CAL ≥3 mm for individuals 1-10 81 43.9 (5.5) Ref. 14-29 years old (Susin et al. 2011) and CAL ≥5 mm for individuals 11-20 113 41.5 (4.6) .75 30 years and over (Susin et al. 2004). Experience of dental caries at 21-28 440 33.2 (2.2) .07 baseline was defined using the median DMFT for the sample (me- BMI status at baseline dian =12). Individuals were categorized according to the number of Normal 312 32.3 (2.7) Ref. teeth present at baseline as follows: 1-10, 11-20, 21-28. Overweight 187 34.2 (3.5) .67 Obese 136 47.1 (4.3) .004 BMI status change (n = 609) 2.7 | Statistical analyses Remained normal 234 31.8 (3.2) Ref. Gained weight 139 32.4 (3.9) .91 Data analysis was performed using a statistical package (Stata Remained overweight 116 36.7 (4.6) .38 version 14 for Macintosh, Stata Corp). Comparisons of incidence Remained obese 120 44.6 (4.5) .02 rates of tooth loss were carried out using the Wald test, with Abbreviation: SE: standard error. alpha set at 5%. Standard errors (SE) were adjusted for multiple aWald test. comparisons. VALLIM et al. | 19

FIGURE 2 Incidence of tooth loss according to BMI and sex (a), presence of periodontitis (b) and smoking exposure (c)

A modified Poisson regression approach to binary outcomes (Zou The mean number of teeth present at baseline was 21.7 (95%CI 2004) was used to estimate the association between BMI, change 21.1-22.2). Overall, 229 individuals lost at least one tooth, resulting in obese status and tooth loss. This approach allows the calcula- in an incidence of tooth loss equal to 36.1% (95%CI 32.3-39.8). There tion of relative risk (RR) in cohort studies. Thus, crude and adjusted were no significant differences in the incidence of tooth loss related RRs and their 95% confidence intervals (CI) were reported. Before to sex or diabetes (Table 1). Older individuals presented a borderline modelling BMI, change in obesity status and tooth loss, univariable significantly higher incidence of tooth loss. The incidence was signifi- models between other exposures and tooth loss were fitted. A final cantly higher among individuals from low education and socio-eco- multivariable model was achieved according to the purposeful ap- nomic statuses compared to the high categories. Heavy smokers had proach (Hosmer & Lemeshow 2000). In summary, variables showing an incidence of 49.2% compared to 33.1% for never-smokers. Higher associations with p values ≤.25 in the univariable models were in- incidence was also observed for individuals with no dental care over cluded in the multivariable model. Maintenance of variables in the the follow-up period, for those with periodontitis and with higher final model was determined by a combination of p values <.05, con- dental caries experience at baseline. Individuals with less teeth at founding and interactions. Thereafter, BMI and change in obesity baseline tended to present higher incidence of tooth loss, but with- status were included separately in the model, and analyses of con- out significant difference. Obese individuals presented significantly founding and interactions were conducted once again. Those expo- higher incidence of tooth loss (47.1%) than normal-weight individuals sures with non-significant associations (p ≥ .05) and demonstrating (32.4%) (p = .004). Individuals that remained obese had significantly no confounding modification defined by at least 25% change in the higher tooth loss (incidence=44.6%) than those that remained nor- estimate of other exposures were excluded from the final multivari- mal weight over the 5 years. able model. Effect modification for the association between tooth loss and baseline BMI was observed for sex, presence of periodontitis at baseline and smoking exposure (Figure 2). The incidence was sig- 3 | RESULTS nificantly higher among obese females than normal-weight females; however, this difference was not significant for males (Figure 2A). At baseline, 1,586 subjects were clinically examined (Figure 1). In The incidence of tooth loss was significantly higher for obese (67.8%) the 5-year follow-up, 755 (47.6%) individuals were re-examined and compared to normal-weight (47.8%) individuals that had periodonti- interviewed. Baseline characteristics of those individuals that par- tis but not for those without periodontitis (32.5% vs. 27.9%) at base- ticipated and not participated in the follow-up examination were line (Figure 2B). The same occurred for ever-smokers compared to compared to respondents. There were no significant differences never-smokers (Figure 2C). between them in regard to socio-economic status, education and Univariable Poisson models for the association between risk smoking habits. Non-participants were slightly younger (36.5 ± 17.6 factors with tooth loss are presented as a supplementary file vs. 39.7 ± 17.0 years) and comprised higher percentages of females (Appendix Table S1). The final multivariable model for exposure (57.5% vs. 51.9%) than participants. There was no significant dif- variables associated with tooth loss included age, socio-economic ference in the number of lost teeth at baseline between non-par- status, smoking exposure, dental care over 5 years, number of pres- ticipants and participants (7.9 ± 8.7 and 7.5 ± 9.3, p = 0.4). Among ent teeth at baseline and periodontitis at baseline. Obese individu- participants, 120 individuals were excluded due to different reasons, als had 35% higher risk for tooth loss than normal-weight individuals resulting in a sample of 635 individuals for the analyses of baseline (Table 2). This association was significant for females (RR=1.48), BMI (Figure 1). Only 26 individuals lost weight over 5 years; thus, but not for males. The stratification for periodontal status at base- they were removed from these analyses because no reliable esti- line showed 51% higher risk for tooth loss in obese (RR=1.51) than mates could be obtained. After their exclusion, 609 individuals pro- normal-weight individuals only for those with periodontitis. The vided data for the analyses of change in obesity status. association between obesity and tooth loss was significant among 20 | VALLIM et al.

TABLE 2 Multivariable Poisson RR 95% CI p regression models of the effect of obesity Whole samplea on tooth loss for the whole sample and Normal 1 according to sex, smoking exposure and Overweight 0.97 0.75-1.26 .86 periodontitis Obese 1.35 1.06-1.72 .01 Stratification for sex Malesa Normal 1 Overweight 0.76 0.51-1.14 .18 Obese 1.22 0.87-1.71 .24 Femalesa Normal 1 Overweight 1.15 0.83-1.58 .41 Obese 1.48 1.08-2.02 .01 Stratification for periodontitis No periodontitisb Normal 1 Overweight 0.91 0.63-1.30 .61 Obese 1.20 0.83-1.73 .33 Periodontitisb Normal 1 Overweight 1.08 0.77-1.52 .81 Obese 1.51 1.12-2.04 .01 Stratification for smoking Never-smokersc Normal 1 Overweight 0.91 0.64-1.29 .60 Obese 1.18 0.82-1.68 .37 Ever-smokersc Normal 1 Overweight 1.04 0.72-1.49 .99 Obese 1.53 1.13-2.08 .01 aAdjusted for age, socio-economic status, smoking exposure, dental care over 5 years, number of teeth at baseline and periodontitis at baseline. bAdjusted for age, socio-economic status, smoking exposure, dental care over 5 years and number of teeth at baseline. cAdjusted for age, socio-economic status, dental care over 5 years, number of teeth at baseline and periodontitis at baseline.

FIGURE 3 Incidence of tooth loss according to change in obesity status and sex (a), presence of periodontitis (b) and smoking exposure (c) VALLIM et al. | 21 ever-smokers, whereas among never-smokers there was no signif- TABLE 3 Final multivariable Poisson models for tooth loss and change in obesity status icant association. All individuals that were obese at baseline remained obese after RR 95% CI p the follow-up period (Appendix Table S2). Among overweight indi- Whole samplea viduals, 65.5% remained overweight and 34.5% gained weight and Remained normal 1 became obese. Among normal-weight individuals, 25.0% gained Gained weight 0.94 0.69-1.27 .67 weight. The incidence of tooth loss according to changes in obesity Remained overweight 1.02 0.74-1.39 .99 status stratified by sex, periodontitis and smoking is presented in the Figure 3. The association between changes in obesity status Remained obese 1.29 0.97-1.71 .07 and tooth loss was not significant in the final multivariable model Stratification for sex a (Table 3). However, there was an increased risk for tooth loss for Males those who remained obese than those who remained normal Remained normal 1 weight among females, those with periodontitis at baseline and Gained weight 0.73 0.46-1.15 .18 ever-smokers. Remained overweight 0.84 0.52-1.35 .47 Remained obese 1.02 0.67-1.54 .93 Femalesa 4 | DISCUSSION Remained normal 1 Gained weight 1.18 0.77-1.80 .44 In the present study, approximately one third of the individuals lost Remained overweight 1.23 0.77-1.94 .36 at least one tooth over 5 years. Obesity was associated with higher Remained obese 1.62 1.10-2.37 .01 incidence of tooth loss. Sex, periodontal status and smoking were Stratification for periodontitis found to be modifiers of this association; females, smokers and indi- b viduals with periodontitis having higher risk than their counterparts. No periodontitis Individuals that remained obese over time had higher risk for tooth Remained normal 1 loss than those that remained normal weight among females, those Gained weight 0.89 0.59-1.38 .63 with periodontitis and ever-smokers. Remained overweight 1.01 0.62-1.64 .95 Previous observational studies with cross-sectional designs Remained obese 1.25 0.81-1.94 .31 have found higher occurrence of tooth loss in obese than in nor- Periodontitisb mal-weight individuals. The estimated odds for tooth loss among Remained normal 1 obese, compared to normal-weight individuals, in these cross-sec- Gained weight 0.97 0.63-1.49 .96 tional studies was approximately 50% higher (Nascimento et al. Remained overweight 1.04 0.69-1.57 .69 2016). More recently, data from NHANES 1999-2012 were anal- Remained obese 1.43 1.04-2.10 .04 ysed demonstrating that central obesity measured by waist–hip Stratification for smoking ratio was associated with prevalence ratio of tooth loss 40% higher Never-smokersc among individuals with BMI <30 kg/m2 (Kang et al. 2019). These findings were corroborated by our study in that the risk for tooth Remained normal 1 loss over 5 years was approximately 30% higher in obese compared Gained weight 0.91 0.56-1.46 .69 to normal-weight individuals. Remained overweight 0.86 0.53-1.40 .55 Tooth loss is the complex outcome between oral disease/con- Remained obese 1.11 0.69-1.77 .67 ditions and patient care/treatment decisions. Systematic reviews Ever-smokersc frequently point out the lack of information about tooth loss in the Remained normal 1 literature, mainly from cohort studies, while highlighting that tooth Gained weight 0.90 0.59-1.38 .69 loss is considered an important part of the global burden of oral con- Remained overweight 1.10 0.71-1.68 .66 ditions. To the best of our knowledge, only one longitudinal study Remained obese 1.45 1.01-2.06 .04 evaluated the impact of obesity on tooth loss (Meisel et al. 2014). aAdjusted for age, socio-economic status, smoking exposure, dental The Study of Health in Pomerania provided data for tooth loss also care over 5 years, number of teeth at baseline and periodontitis at over 5 years, which was correlated with obesity and systemic in- baseline. flammation for the whole sample. However, they found a significant bAdjusted for age, socio-economic status, smoking exposure, dental association between obesity defined by BMI and higher tooth loss care over 5 years and number of teeth at baseline. c only in men after adjustments for important confounders, including Adjusted for age, socio-economic status, dental care over 5 years, number of teeth at baseline and periodontitis at baseline. C-reactive protein. The associations between waist–hip ratio and 22 | VALLIM et al. tooth loss were significant for both sexes. Overall, their findings loss, low socio-economic status and smoking were confirmed. corroborate the main associations observed in this study, in which Interestingly, individuals reporting higher frequency of dental obesity is a risk factor for tooth loss. care over the follow-up time had higher incidence of tooth loss It has been demonstrated in the literature that an effect mod- than those not visiting the dentist in the same period. It would be ification of sex in different chronic diseases may exist due to sex expected that visiting a dentist would prevent tooth loss (Foiles differences related to obesity, including anatomical adipose tissue Sifuentes et al. 2020; Li et al. 2011). However, other studies have distribution, sex hormone effects, receptor activity, genetic influ- shown associations similar to that of the present study (Bhat et al. ences and inflammatory responses (Moreira-Pais et al. 2020; Palmer 2018; Gilbert et al. 2003), which may be explained in part by so- & Clegg 2015). In this study, obesity was a risk factor for tooth loss cial disparities, reasons to consult a dentist and by the fact that among females, but not among males. In a previous analysis of the the majority of tooth extractions are performed in dental offices/ same sample, it was demonstrated that obesity was also associ- services. ated with higher progression of periodontitis only in females (Gaio The population-based sample, 5 years of follow-up, evaluation et al. 2016). This may be one possible explanation for the present of changes in BMI status and assessment of effect modification are findings, since periodontitis was related to tooth loss in these anal- among the major strengths of this study. One limitation of this study yses. It should be also acknowledged that BMI has limitations to is the lack of assessment of biomarkers for systemic inflammation. differentiate body fat composition and distribution between sexes BMI has limitations in terms of body fat determination (Meeuwsen (Meeuwsen et al. 2010). et al. 2010), although it is recommended by WHO, since it is easy to Dental caries and periodontitis are the major causes of tooth be applied in epidemiological studies and it is highly correlated with loss. However, the impact of each disease may be different in dif- major chronic diseases (Gutin 2018). ferent populational groups (Eklund & Burt 1994; Milgrom & Reisine It can be concluded that obesity is a risk factor for tooth loss. 2000). In this study, caries and periodontitis at baseline were asso- Females, individuals with periodontitis and smokers deserve major ciated with higher tooth loss in the crude analyses; however, only attention in preventive strategies for tooth loss due to negative im- periodontitis remained in the final multivariable model confirming pacts of obesity on tooth retention over time in these subgroups. previous findings that periodontitis leads to tooth loss. Additionally, Weight loss or gain still needs to be studied in the future to deter- periodontitis modified the association between obesity and tooth mine their impact on incident tooth loss. loss, indicating a greater negative impact of this disease on tooth retention. Previous studies have also found no effect of caries on mod- CONFLICT OF INTEREST els for tooth loss (Meisel et al. 2014; Pilotto et al. 2014). It should The authors declare no conflict of interest associated with the pre- be acknowledged that the present study did not address possible sent study. reasons for tooth loss. Obesity at baseline was associated with higher risk for tooth loss ORCID in ever-smokers, but not in never-smokers in this study. This may Jasim M. Albandar https://orcid.org/0000-0001-7801-3811 be a consequence of increased risk for periodontitis among smok- Alex Nogueira Haas https://orcid.org/0000-0003-0531-6234 ers. Noteworthy, interaction between smoking and obesity parameters has also been observed for the risk for cardiovascular disease REFERENCES (Luo et al. 2019) and rheumatoid arthritis (Hedstrom et al. 2019). Bernardo Cde, O., Boing, A. F., Vasconcelos Fde, A., Peres, K. G., & Although there might be a synergistic effect of smoking and obesity Peres, M. A. (2012). Association between tooth loss and obesity in Brazilian adults: a population-based study. Rev Saude Publica, on immune and inflammatory responses, evidence is still needed to 46(5), 834-842. https://doi.org/10.1590/s0034-8910201200 elucidate these events regarding tooth loss. 0500010 The associations observed between baseline BMI and tooth loss Bhat, M., Do, L. G., & Roberts-Thomson, K. (2018). Association between were stronger than those found between changes in obesity status dental visiting and missing teeth: Estimation using propensity score adjustment. Journal of Investigative and Clinical Dentistry, 9(3), and tooth loss. For example, it would be expected that losing weight e12326. https://doi.org/10.1111/jicd.12326 and becoming normal weight would lead to lower tooth loss; never- Bluher, M. (2019). Obesity: global epidemiology and pathogenesis. Nature theless, this could not be assessed in this study due to the low num- Reviews Endocrinology, 15(5), 288-298. https://doi.org/10.1038/ ber of individuals presenting this condition. Significant associations s41574-019-0176-8 were only found for females, those with periodontitis at baseline and Collaborators, G. B. D. O. D., Bernabe, E., Marcenes, W., Hernandez, C. R., Bailey, J., Abreu, L. G., … Kassebaum, N. J. (2020). Global, Regional, ever-smokers, but not for the whole sample after statistical adjust- and National Levels and Trends in Burden of Oral Conditions ments. This may be due to lack of power of the sample since most p from 1990 to 2017: A Systematic Analysis for the Global Burden values were borderline. of Disease 2017 Study. Journal of Dental Research, 99(4), 362-373. In this study, the associations between obesity and tooth loss https://doi.org/10.1177/0022034520908533 Dye, B. A., Weatherspoon, D. J., & Lopez Mitnik, G. (2019). Tooth loss were adjusted for important demographic and behavioural risk among older adults according to poverty status in the United factors/indicators previously determined in the literature. In our States. The Journal of the American Dental Association, 150(1), 9-23. analyses, previous associations demonstrated between tooth e3. https://doi.org/10.1016/j.adaj.2018.09.010 VALLIM et al. | 23

Eklund, S. A., & Burt, B. A. (1994). Risk factors for total tooth loss edentulism: a marginal approach. BMC Oral Health, 11, 9. https:// in the United States; longitudinal analysis of national data. doi.org/10.1186/1472-6831-11-9 Journal of Public Health Dentistry, 54(1), 5-14. https://doi. Luo, W. S., Chen, F., Ji, J. M., & Guo, Z. R. (2019). Interaction of tobacco org/10.1111/j.1752-7325.1994.tb01173.x smoking and alcohol consumption with obesity on cardiovascular Ellulu, M. S., Patimah, I., Khaza'ai, H., Rahmat, A., & Abed, Y. (2017). disease in a Chinese cohort. Coronary Artery Disease, 31(4), 372- Obesity and inflammation: The linking mechanism and the com- 377. https://doi.org/10.1097/MCA.0000000000000837 plications. Archives of Medical Science, 13(4), 851-863. https://doi. Meeuwsen, S., Horgan, G. W., & Elia, M. (2010). The relationship be- org/10.5114/aoms.2016.58928 tween BMI and percent body fat, measured by bioelectrical imped- Emami, E., de Souza, R. F., Kabawat, M., & Feine, J. S. (2013). The impact ance, in a large adult sample is curvilinear and influenced by age and of edentulism on oral and general health. International Journal of sex. Clinical Nutrition, 29(5), 560-566. https://doi.org/10.1016/j. Dentistry, 2013, 1-7. https://doi.org/10.1155/2013/498305 clnu.2009.12.011 Foiles Sifuentes, A. M., Castaneda-Avila, M. A., & Lapane, K. L. (2020). Meisel, P., Holtfreter, B., Volzke, H., & Kocher, T. (2014). Sex differences of The relationship of aging, complete tooth loss, and having a dental tooth loss and obesity on systemic markers of inflammation. Journal visit in the last 12 months. Clin Exp Dent Res, 6, 550-557. https://doi. of Dental Research, 93(8), 774-779. https://doi.org/10.1177/00220 org/10.1002/cre2.309 34514535604 Gaio, E. J., Haas, A. N., Rosing, C. K., Oppermann, R. V., Albandar, J. M., Milgrom, P., & Reisine, S. (2000). Oral health in the United States: the & Susin, C. (2016). Effect of obesity on periodontal attachment loss post-fluoride generation. Annual Review of Public Health, 21, 403- progression: a 5-year population-based prospective study. Journal 436. https://doi.org/10.1146/annurev.publhealth.21.1.403 of Clinical Periodontology, 43(7), 557-565. https://doi.org/10.1111/ Moreira-Pais, A., Ferreira, R., Neves, J. S., Vitorino, R., Moreira- jcpe.12544 Goncalves, D., & Nogueira-Ferreira, R. (2020). Sex differences on Gilbert, G. H., Duncan, R. P., & Shelton, B. J. (2003). Social determinants adipose tissue remodeling: from molecular mechanisms to thera- of tooth loss. Health Services Research, 38(6 Pt 2), 1843-1862. peutic interventions. Journal of Molecular Medicine, 98(4), 483-493. https://doi.org/10.1111/j.1475-6773.2003.00205.x https://doi.org/10.1007/s00109-020-01890-2 Gutin, I. (2018). In BMI We Trust: Reframing the Body Mass Index as a Muller, F., Naharro, M., & Carlsson, G. E. (2007). What are the prevalence Measure of Health. Social Theory & Health, 16(3), 256-271. https:// and incidence of tooth loss in the adult and elderly population in doi.org/10.1057/s41285-017-0055-0 Europe? Clinical Oral Implants Research, 18(Suppl. 3), 2-14. https:// Haas, A. N., Gaio, E. J., Oppermann, R. V., Rosing, C. K., Albandar, doi.org/10.1111/j.1600-0501.2007.01459.x J. M., & Susin, C. (2012). Pattern and rate of progression Nascimento, G. G., Leite, F. R., Conceicao, D. A., Ferrua, C. P., Singh, A., of periodontal attachment loss in an urban population of & Demarco, F. F. (2016). Is there a relationship between obesity and South Brazil: a 5-years population-based prospective study. tooth loss and edentulism? A systematic review and meta-analysis. Journal of Clinical Periodontology, 39(1), 1-9. https://doi. Obesity Reviews, 17(7), 587-598. https://doi.org/10.1111/obr.12418 org/10.1111/j.1600-051X.2011.01818.x. Nascimento, G. G., Leite, F. R., Do, L. G., Peres, K. G., Correa, M. B., Haslam, D. W., & James, W. P. (2005). Obesity. The Lancet, 366(9492), Demarco, F. F., & Peres, M. A. (2015). Is weight gain associated with 1197-1209. https://doi.org/10.1016/S0140-6736(05)67483-1 the incidence of periodontitis? A systematic review and meta-anal- Haworth, S., Shungin, D., Kwak, S. Y., Kim, H. Y., West, N. X., Thomas, ysis. Journal of Clinical Periodontology, 42(6), 495-505. https://doi. S. J., … Johansson, I. (2018). Tooth loss is a complex measure of org/10.1111/jcpe.12417 oral disease: Determinants and methodological considerations. Ojima, M., Hanioka, T., Tanaka, K., & Aoyama, H. (2007). Cigarette Community Dentistry and Oral Epidemiology, 46(6), 555-562. https:// smoking and tooth loss experience among young adults: a na- doi.org/10.1111/cdoe.12391 tional record linkage study. BMC Public Health, 7, 313. https://doi. Hedstrom, A. K., Klareskog, L., & Alfredsson, L. (2019). Interplay be- org/10.1186/1471-2458-7-313 tween obesity and smoking with regard to RA risk. RMD Open, 5(1), Ostberg, A. L., Nyholm, M., Gullberg, B., Rastam, L., & Lindblad, U. e000856. https://doi.org/10.1136/rmdopen-2018-000856 (2009). Tooth loss and obesity in a defined Swedish population. Hosmer, D. W., & Lemeshow, S. L. (2000). Applied logistic regression (3rd Scandinavian Journal of Public Health, 37(4), 427-433. https://doi. ed.). John Wiley & Sons. org/10.1177/1403494808099964 Jiang, Y., Okoro, C. A., Oh, J., & Fuller, D. L. (2013). Sociodemographic Palmer, B. F., & Clegg, D. J. (2015). The sexual dimorphism of obesity. and health-related risk factors associated with tooth loss among Molecular and Cellular Endocrinology, 402, 113-119. https://doi. adults in Rhode Island. Preventing Chronic Disease, 10, E45. https:// org/10.1016/j.mce.2014.11.029 doi.org/10.5888/pcd10.110285 Pilotto, L. M., Celeste, R. K., Faerstein, E., & Slavutzky, S. M. (2014). Jung, S. H., Ryu, J. I., & Jung, D. B. (2011). Association of total tooth Association between tooth loss and overweight/obesity among loss with socio-behavioural health indicators in Korean el- Brazilian adults: the Pro-Saude Study. Brazilian Oral Research, 28, derly. Journal of Oral Rehabilitation, 38(7), 517-524. https://doi. https://doi.org/10.1590/1807-3107bor-2014.vol28.0032 org/10.1111/j.1365-2842.2010.02178.x Seerig, L. M., Nascimento, G. G., Peres, M. A., Horta, B. L., & Demarco, Kang, J., Smith, S., Pavitt, S., & Wu, J. (2019). Association between cen- F. F. (2015). Tooth loss in adults and income: Systematic review and tral obesity and tooth loss in the non-obese people: Results from meta-analysis. Journal of Dentistry, 43(9), 1051-1059. https://doi. the continuous National Health and Nutrition Examination Survey org/10.1016/j.jdent.2015.07.004 (NHANES) 1999–2012. Journal of Clinical Periodontology, 46(4), Singh, A., Peres, M. A., Peres, K. G., Bernardo Cde, O., Xavier, A., & 430-437. https://doi.org/10.1111/jcpe.13091 D'Orsi, E. (2015). Gender differences in the association be- Kassebaum, N. J., Bernabe, E., Dahiya, M., Bhandari, B., Murray, C. J., tween tooth loss and obesity among older adults in Brazil. & Marcenes, W. (2014). Global Burden of Severe Tooth Loss: A Revista de Saúde Pública, 49, 44. https://doi.org/10.1590/S0034- Systematic Review and Meta-analysis. Journal of Dental Research, 8910.2015049005590 93(7 suppl), 20S-28S. https://doi.org/10.1177/0022034514537828 Singh, G. M., Danaei, G., Farzadfar, F., Stevens, G. A., Woodward, M., Kopelman, P. G. (2000). Obesity as a medical problem. Nature, 404(6778), Wormser, D., … Prospective Studies, C. (2013). The age-specific 635-643. https://doi.org/10.1038/35007508 quantitative effects of metabolic risk factors on cardiovascular Li, K. Y., Wong, M. C., Lam, K. F., & Schwarz, E. (2011). Age, pe- diseases and diabetes: a pooled analysis. PLoS ONE, 8(7), e65174. riod, and cohort analysis of regular dental care behavior and https://doi.org/10.1371/journal.pone.0065174 24 | VALLIM et al.

Susin, C., Dalla Vecchia, C. F., Oppermann, R. V., Haugejorden, O., & Zou, G. (2004). A modified poisson regression approach to prospective Albandar, J. M. (2004). Periodontal attachment loss in an urban studies with binary data. American Journal of Epidemiology, 159(7), population of Brazilian adults: effect of demographic, behavioral, 702-706. and environmental risk indicators. Journal of Periodontology, 75(7), 1033-1041. Susin, C., Haas, A. N., Valle, P. M., Oppermann, R. V., & Albandar, SUPPORTING INFORMATION J. M. (2011). Prevalence and risk indicators for chronic peri- Additional supporting information may be found online in the odontitis in adolescents and young adults in south Brazil. Journal of Clinical Periodontology, 38(4), 326-333. https://doi. Supporting Information section. org/10.1111/j.1600-051X.2011.01699.x Susin, C., Oppermann, R. V., Haugejorden, O., & Albandar, J. M. (2005). Tooth loss and associated risk indicators in an adult urban popula- How to cite this article: Vallim AC, Gaio EJ, Oppermann RV, tion from south Brazil. Acta Odontologica Scandinavica, 63(2), 85- et al. Obesity as a risk factor for tooth loss over 5 years: A 93. https://doi.org/10.1080/00016350510019694 population-based cohort study. J Clin Periodontol. Vazquez-Nava, F., Vazquez-Rodriguez, E. M., Saldivar-Gonzalez, A. H., Lin-Ochoa, D., Martinez-Perales, G. M., & Joffre- 2021;48:15–24. https://doi.org/10.1111/jcpe.13378 Velazquez, V. M. (2010). Association between obesity and dental caries in a group of preschool children in Mexico. Journal of Public Health Dentistry, 70(2), 124-130. https://doi. org/10.1111/j.1752-7325.2009.00152.x Received: 23 January 2020 | Revised: 2 October 2020 | Accepted: 6 October 2020 DOI: 10.1111/jcpe.13388

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

Association of hyperglycaemia with periodontal status: Results of the Northern Finland Birth Cohort 1966 study

1Research Unit of Oral Health Sciences, Faculty of Medicine, University of Oulu, Abstract Oulu, Finland Aim: To investigate the association of hyperglycaemia and changes in glycaemic con- 2Department of Oral and Maxillofacial Surgery, Oulu University Hospital, Oulu, trol with periodontal status in non-diabetic individuals. Finland Materials and methods: A sub-population (n = 647) of the Northern Finland Birth 3 Center for Life Course Epidemiology and Cohort 1966 was studied. We categorized long-term glucose balance based on fast- Systems Medicine, University of Oulu and Unit of Primary Care, Oulu, Finland ing plasma glucose (FPG) at ages 31 and 46: FPG <5.0 mmol/l (strict normoglycaemia), 4Medical Research Center, Oulu University FPG 5.0–5.59 mmol/l (slightly elevated FPG) and FPG 5.6–6.9 mmol/l (prediabetes). Hospital and University of Oulu, Oulu, Probing pocket depth (PPD) and alveolar bone level (BL) data were collected at age Finland 5Healthcare and Social Services of 46. Relative risks (RR, 95% CI) were estimated using Poisson regression models. Selänne, Pyhäjärvi, Finland Results: Periodontal status was poorer in individuals whose glucose balance wors- 6 Center for Life Course Health Research, ened from age 31 to 46 years than in those with a stable glucose balance. In the case Faculty of Medicine, University of Oulu, Health Centre of Oulu, Oulu, Finland of strict normoglycaemia at age 31 and slightly elevated FPG or prediabetes at age 46, 7Oulunkaari Health Centre, Ii, Finland the RRs for PPD ≥4 mm were 1.8 (95% CI 1.4–2.2) and 2.8 (95% CI 2.0–3.8) and for BL ≥5 mm 1.1 (95% CI 0.8–1.4) and 1.8 (95% CI 1.2–2.8), respectively. Correspondence Paula Tegelberg, Research Unit of Oral Conclusion: The results of this population-based cohort study suggest that impair- Health Sciences, University of Oulu, PO Box 5000, 90014 University of Oulu, ment in glucose control in non-diabetic individuals is associated with periodontal Oulu, Finland. pocketing and alveolar bone loss. Email: [email protected]

KEYWORDS alveolar bone loss, hyperglycaemia, periodontal pocket, periodontitis, prediabetes

1 | INTRODUCTION by American Diabetes Association (ADA) (Yip et al., 2017), it has been reported to affect 48.1% of the Asian population and 58.9% of Prediabetes denotes the precursor stage of diabetes mellitus char- European Caucasian adults. acterized by fasting plasma glucose levels which are elevated but From the point of view of public health, it is important to keep below the threshold for the diagnosis of diabetes. Prediabetes has in mind that up to 70% of individuals with prediabetes have been been defined using several measures, including impaired fasting estimated to develop diabetes (Nathan et al., 2007). The clinical glucose, impaired glucose tolerance and elevated glycated hae- complications of diabetes mellitus include micro- and macrovascular moglobin (American Diabetes Association, 2017). The prevalence complications such as cardiovascular disease, retinopathy, neuropa- of prediabetes is dependent on which diagnostic test is used, and thy and nephropathy. The underlying mechanisms in the pathogen- population characteristics such as age and gender distribution, and esis of these complications are that subjects with hyperglycaemia ethnic composition. Prediabetes is common; based on the definition have multiple biological abnormalities, such as an imbalance of

wileyonlinelibrary.com/journal/jcpe | 25 J Clin Periodontol. 2021;48:25–37. 26 | TEGELBERG et al. cytokines and adipokines, oxidative stress and accumulation of advanced glycation end products (AGEs) and their receptor RAGE Clinical Relevance in tissues (Giacco & Brownlee, 2010; King, 2008). Subjects with Scientific rationale for the study: A number of cross-sec- hyperglycaemia also have impaired tissue repair and poor wound tional studies show that prediabetes is associated with healing (Baltzis et al., 2014). Many of the above biological effects of periodontitis but information based on longitudinal studies hyperglycaemia are also observable in periodontal tissues (Polak & is scarce. Moreover, the effects of early hyperglycaemia Shapira, 2018) and, in fact, periodontitis has been declared the "sixth under the threshold for prediabetes on the periodontium complication" of diabetes mellitus (Löe, 1993). are largely unknown. In view of the overall high prevalence of both prediabetes Principal findings: This study provided evidence that pre- and periodontal infection, it is important to study whether the diabetes and even slightly elevated fasting plasma glu- risk for periodontal tissue destruction is increased in non-dia- cose levels may be associated with ≥4 mm probing pocket betic individuals with elevated fasting plasma glucose. Most of depths. the earlier studies are cross-sectional and report positive associ- Practical implications: Future studies using more accurate ations between prediabetes and periodontitis (Hong et al., 2016; monitoring of dysglycaemia are needed to confirm the as- Lamster et al., 2014; Pérez et al., 2017). To our knowledge, only sociation of early hyperglycaemia with periodontal status. two longitudinal studies exist in this area; one of them reported an increased incidence of periodontal infection in prediabetic individuals (Chiu et al., 2015) and the other one in individuals with insulin resistance/disturbed beta cell function (Timonen 2.1 | Periodontal examination et al., 2013). The strength of these longitudinal studies is that they took into account the temporal sequence between the peri- Periodontal data, including dental plaque, probing pocket depth odontal and the dysglycaemic conditions. In this study, we hy- (PPD) and bleeding on probing (BOP) from all teeth excluding third pothesized that fasting plasma glucose level is associated with molars and residual roots were collected by seven calibrated den- periodontal status in non-diabetic individuals. Subsequently, we tists using a mouth mirror and a ball-pointed periodontal probe investigated whether the changes in fasting plasma glucose level with 2 mm graduations (LM 8-520B, Lääkintämuovi, Finland). between ages 31 and 46 were associated with the number of Before conducting the study, the examiners underwent training on sites with ≥4 mm probing pocket depth and ≥5 mm alveolar bone periodontal examination in volunteer subjects under the supervi- level at age 46. sion of an experienced periodontist (TT). The training included a calibration of probing pressure of 25 g using a letter scale. The examiners also practiced the application of periodontal measurement 2 | MATERIALS AND METHODS criteria in a group of 12 individuals. Immediate feedback was given to the questions raised by the examiners in these exercises and The Northern Finland Birth Cohort 1966 (NFBC1966, University also to the questions/inquiries that were entered in writing in the of Oulu 1966) is a cohort study that examined the health and well- shared research diary during the field work. Replicate periodontal being of individuals born in 1966 in the two northernmost provinces measurements in the cohort subjects were used to calculate the in Finland (Oulu and Lapland). In the 2012–2013 data collection, agreement percentages and the inter-examiner and intra-examiner when the subjects were 46-year-olds, an oral health examination Cohen's kappa values (95% CI) in 108 and 129 individuals, respec- was conducted for the first time on 1,964 participants living in the tively. The examiner with the most clinical experience served as the city of Oulu or within a 100 km radius of Oulu. Prior to that, gen- "gold standard" examiner. eral health data were gathered 15 years earlier when the cohort was The absence/presence of plaque was examined from buccal 31 years old. In this study, we used the periodontal data from the surfaces of the teeth (single score per tooth) and categorized as examination at age 46 and the fasting plasma glucose (FPG) data follows: 1 = no plaque, 2 = visible plaque or plaque, as measured from the examinations at ages 31 and 46. Subjects with manifest by touching the tooth surface lightly with the periodontal probe diabetes mellitus (DM 1 and 2), rheumatic and inflammatory intesti- (Table 3a). nal and lung diseases were excluded; these diagnoses were obtained Probing pocket depths (PPD) were measured at four sites per from the registries of the Finnish Institute for Health and Welfare tooth (mesiobuccal, midbuccal, distobuccal, midoral). Probing force (THL). In addition, individuals with DM 2 screened in the oral glu- (25 g) was calibrated using a letter scale before examining each par- cose tolerance test of the age 46 follow-up study were excluded. ticipant. The inter- and intra-examiner agreements for PPD mea- Data on smoking history were collected using a questionnaire, and surements were 70% and 81%, and the Cohen's kappa values were only never-smokers were included in the current analyses (n = 647). 0.39 (95% CI 0.34–0.43) and 0.61 (95% CI 0.58–0.64), respectively. The study was approved by the Ethical Committee of the Hospital Bleeding on periodontal probing (BOP) at the four sites was reg- District of Northern Ostrobothnia. istered (Table 3a). TEGELBERG et al. | 27

2.2 | Radiographic examination • FPG <5.6 mmol/l at ages 31 and 46, n = 440 (reference group), • FPG <5.6 mmol/l (normoglycaemia) at 31 years and FPG 5.6– Alveolar bone level (BL) on approximal sites of all teeth was meas- 6.9 mmol/l (prediabetes) at age 46, n = 149, ured using panoramic radiographs by a dental hygienist. Panoramic radiographs were available of 568 never-smoking subjects. The mean The three-class categorization: number (± SD) of measurable sites per individual was 33.0 (± 9.1). The distance from the cementoenamel junction (CEJ) to the point • FPG <5.0 mmol/l at ages 31 and 46, n = 109 (reference group), where the periodontal ligament space still remained at its normal • FPG <5.0 mmol/l (strict normoglycaemia) at age 31 and FPG 5.0– width was measured in millimetres. Any sites where the CEJ or the 5.59 mmol/l (slightly elevated FPG) at age 46, n = 179, alveolar crest could not be identified were excluded. Reproducibility • FPG <5.0 mmol/l (strict normoglycaemia) at age 31 and FPG 5.6– was assessed by replicate measurements using the radiographs of 6.9 mmol/l (prediabetes) at age 46, n = 56. the cohort participants. The supervising periodontist (TT) served as the reference examiner. The inter- and intra-examiner agreements for BL measurements were 67% and 70% and the Cohen's kappa 2.4 | Lipid profile values were 0.54 (95% CI 0.50–0.57) and 0.59 (95% CI 0.54–0.63), respectively. Total cholesterol, triglycerides and high-density lipoprotein (HDL) were measured at age 46 from the subjects’ venous blood samples and used in the analyses as continuous variables (mmol/l) (Tables 2.3 | Fasting plasma glucose (FPG) 1–3).

Determination of the glucose balance was based on the FPG levels in the subjects’ venous blood samples. We used two parallel criteria 2.5 | Other variables to determine different FPG categories separately at ages 31 and 46. First, we categorized the subjects into two groups based on the The data on socio-demographics and health behaviour were based ADA criteria; those with normoglycaemia (FPG <5.6 mmol/l) and on a questionnaire at age 46. Education was categorized into three those with prediabetes (FPG 5.6–6.9 mmol/l) (American Diabetes categories: basic (vocational school or lower), intermediate (gradu- Association, 2017). Next the normoglycaemic individuals were fur- ated from a vocational college or university of applied sciences) ther categorized into two groups: individuals with FPG <5.0 mmol/l and higher (graduated from a university) following the international vs. FPG 5.0–5.59 mmol/l. Subsequently, the following three cat- standard classification of education (ISCED, 1997). The frequency of egories existed: FPG <5.0 mmol/l (strict normoglycaemia), FPG physical exercise was categorized into four categories: 2–3 times a 5.0–5.59 mmol/l (slightly elevated FPG) and FPG 5.6–6.9 mmol/l month or less frequently, once a week, 2–3 times a week and daily (prediabetes). The above FPG categorizations at ages 31 and 46 or 4–6 times a week (Tammelin et al., 2003). The questions of oral were then combined to determine the long-term glucose bal- health habits, toothbrushing frequency and time since the most re- ance—either stable or worsened—and the following two-class or cent dental visit, were based on the guidelines of the World Health three-class categorizations were made for the statistical analyses Organization (2013). In this study, toothbrushing variable was cat- (Table 4). egorized into four categories: twice a day or more often, once a day, Stable glucose balance; an individual stayed in the same FPG cat- and less frequently, and the most recent dental visit into four category at ages 31 and 46. egories: less than 12 months ago, 1 year but less than 2 years ago The two-class categorization (ADA): and 2 years or more ago. Alcohol consumption was used both as a continuous (g/day) and as a categorized variable: abstainers (0 g/day) • FPG <5.6 mmol/l (normoglycaemia), n = 440 (reference group), and tertiles of alcohol users (< 3.0 g/day, 3.0–11.1 g/day and >11.1 g/ • FPG 5.6–6.9 mmol/l (prediabetes), n = 24, day). Body mass index (BMI, kg/m2) was categorized into three categories: <24.9, 25.0–29.9 and ≥30.0 (World Health Organization, The three-class categorization: 2000).

• FPG <5.0 mmol/l (strict normoglycaemia), n = 109 (reference group), 2.6 | Statistical analyses • FPG 5.0–5.59 mmol/l (slightly elevated FPG), n = 119, • FPG 5.6–6.9 mmol/l (prediabetes), n = 24. In the analyses, the numbers of sites with PPD ≥4 mm and BL ≥5 mm were used as outcome variables (count variables). Relative risks (RR) Worsened glucose balance; a shift to poorer glucose balance and 95% confidence intervals (95% CI) were estimated using Poisson from age 31 to 46 occurred: regression models (Table 4). The categories of individuals with FPG The two-class categorization (ADA): <5.6 mmol/l (the two-class categorization) and <5.0 mmol/l (the 28 | TEGELBERG et al.

TABLE 1 Basic characteristics of the study population; proportions/means and their standard deviations (SD) according to the number of probing pocket depths (PPD ≥4 mm)

Number of sites with PPD ≥4 mm

0 1–3 4–7 ≥ 8

n = 369 n = 184 n = 51 n = 35

Gender, % (n = 639) Male 37.4 45.7 41.2 57.1 Female 62.6 54.3 58.8 42.9 Education, % (n = 623) Basic 24.9 32.4 28 54.3 Intermediate 51.7 47.7 54 37.1 Higher 23.5 19.9 18 8.6 Number of teeth, mean (SD) (n = 639) 27.0 (1.4) 27.0 (1.5) 27.0 (1.4) 26.9 (1.5) Number of teeth with plaque, mean (SD) (n = 639) 4.9 (5.7) 5.5 (5.8) 7.8 (7.5) 10.7 (8.9) Toothbrushing frequency, % (n = 638) Twice a day or more 75.8 72.8 72.5 45.7 Once a day 24.2 26.6 27.5 54.3 Less frequently 0 0.5 0 0 Most recent dental visit, % (n = 620) Less than 1 year ago 62.6 58.6 50 38.7 1–2 years ago 27.4 27.6 28 35.5 ≥3 years ago 10.1 13.8 22 25.8 Frequency of physical exercise, % (n = 632) 2–3 times a month or less frequently 18.5 22.8 26 31.4 Once a week 20.2 20 24 14.3 2–3 times a week 43.3 41.7 38 34.3 Daily or 4–6 times a week 18 15.6 112 20 Alcohol consumption, % (n = 639) 0 g/day 13 14 11.8 11.4 0.001–3.0 g/day 39.8 34.8 31.4 40 >3.0–11.1 g /day 28.2 31.5 29.4 25.7 >11.1 g/day 19 19.6 27.5 22.9 BMI (kg/m2), % (n = 639) <24.9 47.2 42.9 39.2 22.7 25–29.9 36 43.5 35.3 40 ≥30 16.8 13.6 25.5 34.3 Total cholesterol (mmol/l), mean (SD) (n = 639) 5.3 (0.8) 5.2 (0.9) 5.3 (1.1) 5.5 (1.0) HDL cholesterol (mmol/l), mean (SD) (n = 639) 1.6 (0.4) 1.6 (0.4) 1.7 (0.5) 1.5 (0.3) Triglycerides (mmol/l), mean (SD) (n = 639) 1.1 (0.6) 1.1 (0.6) 1.1 (0.5) 1.6 (1.3) Stable glucose balance, % (n = 249) FPG <5.0 mmol/l at 31 and 46 years (ref.) 51 29.2 66.7 18.8 FPG 5.0–5.59 mmol/l at 31 and 46 years 40.9 58.3 33.3 68.8 FPG 5.6–6.9 (prediabetes) at 31 and 46 years 8.1 12.5 0 12.5 Worsened glucose balance, % (n = 340) FPG <5.0 mmol/l at 31 and 46 years (ref.) 35.5 24.7 29.6 21.4 FPG <5.0 mmol/l at 31 and 5.0–5.59 mmol/l at 46 years 49.1 56.5 55.6 64.3 FPG <5.0 mmol/l at 31 and 5.6–6.9 (prediabetes) at 46 years 15.4 18.8 14.8 14.3

Abbreviations: BMI, body mass index; HDL, high-density lipoprotein; FPG, fasting plasma glucose. TEGELBERG et al. | 29

TABLE 2 Basic characteristics of the study population; proportions/means and their standard deviations (SD) according to the number of the sites with ≥5 mm alveolar bone level (BL)

Number of sites with BL ≥5 mm

0 1–3 ≥ 4

n = 307 n = 225 n = 36

Gender, % (n = 568) Male 39.1 44.4 47.2 Female 60.9 55.6 52.8 Education, % (n = 553) Basic 26.4 32.7 25.5 Intermediate 48.5 51.4 52.9 Higher 25.1 15.9 20.6 Number of teeth, mean (SD) (n = 566) 27.0 (1.4) 27.0 (1.5) 26.8 (1.7) Number of teeth with plaque, mean (SD) (n = 564) 5.3 (6.1) 5.4 (5.9) 6.3 (6.4) Toothbrushing frequency, % (n = 567) Twice a day or more 74.3 74.7 74.3 Once a day 25.4 25.3 25.7 Less frequently 0.3 0 0 Most recent dental visit, % (n = 549) Less than 1 year ago 62.4 52.1 60 1–2 years ago 24.4 33.8 28.6 ≥3 years ago 13.2 14.2 11.4 Frequency of physical exercise, % (n = 561) 2–3 times a month or less frequently 19.7 23.3 23.5 Once a week 20.1 22.4 11.8 2–3 times a week 44.1 38.1 47.1 Daily or 4–6 times a week 16.1 16.1 17.6 Alcohol consumption, % (n = 568) 0 g/day 15.6 12.4 8.3 0.001–3.0 g/day 38.1 38.7 38.9 >3.0–11.1 g/day 29.6 30.2 25 >11.1 g/day 16.6 18.7 27.8 BMI (kg/m2), % (n = 568) <24.9 44.3 43.1 50 25–29.9 36.8 40.9 27.8 ≥30 18.9 16 22.2 Total cholesterol (mmol/l), mean (SD) (n = 568) 5.3 (0.9) 5.3 (0.8) 5.4 (1.1) HDL cholesterol (mmol/l), mean (SD) (n = 568) 1.6 (0.4) 1.5 (0.4) 1.6 (0.4) Triglycerides (mmol/l), mean (SD) (n = 568) 1.1 (0.6) 1.1 (0.7) 1.3 (1.3) Stable glucose balance, % (n = 227) FPG <5.0 mmol/l at 31 and 46 years (ref.) 43.7 44.3 53.8 FPG 5.0–5.59 mmol/l at 31 and 46 years 45.2 48.9 30.8 FPG 5.6–6.9 (prediabetes) at 31 and 46 years 11.1 6.8 15.4 Worsened glucose balance, % (n = 302) FPG <5.0 mmol/l at 31 and 46 years (ref.) 35 30.5 41.2 FPG <5.0 mmol/l at 31 and 5.0–5.59 mmol/l at 46 years 47.8 57 41.2 FPG <5.0 mmol/l at 31 and 5.6–6.9 (prediabetes) at 46 years 17.2 12.5 17.6

Abbreviations: BMI, body mass index; HDL, high-density lipoprotein; FPG, fasting plasma glucose. 30 | TEGELBERG et al. (Continues) n = 56) 24.1 18.5 40.7 16.7 57.4 31.5 11.1 63.6 34.5 1.8 7.3 32.7 29.1 30.9 <5.0 mmol/l at and 31 5.6–6.9 mmol/l (prediabetes) at 46 years 24.5 45.3 30.2 ( 54.5 45.5 n = 179) 20.8 17.4 43.3 18.5 63.4 25.6 11 78.2 52.7 0 8.9 38.3 36.7 16.1 25.8 52.8 21.3 Worsened glucose balance glucose Worsened <5.0 mmol/l at and 31 5.0– 5.59 mmol/l at 46 years ( 33.3 66.7 = 24) 21.7 21.7 34.8 21.7 45.5 36.4 18.2 62.5 37.5 0 5.6–6.9 mmol/l (prediabetes) at and 31 46 years 8.3 45.8 25 20.8 (n 50 45.8 4.2 79.2 20.8 = 119) 19.8 24.8 41 14.5 59.1 22.6 18.3 68.1 31.9 0 Stable glucose balance glucose Stable 5.0–5.59 mmol/l at and31 46 years 14.3 40.3 24.4 21 (n 28.3 54 17.7 47.9 52.1 344) 337) n = 109) 344) 20.4 17.6 43.5 18.5 58.5 32.1 9.4 85.3 14.7 0 Reference Reference category <5.0 mmol/l at 31 and 46 years 18.3 40.4 32.1 9.2 ( 21.7 50.9 27.4 14.7 85.3 252) (worsened,n = 252) n = g/day frequently g/day 2–3 times2–3 a month or less Once a week times2–3 a week Daily or 4–6 times a week Less than 1 year ago years ago 1–2 years≥3 ago Twice a day orTwice more Once a day frequently Less 0 0.001–3.0 g /day >3.0–11.1 g/day >11.1 Basic Intermediate Higher Male Female Frequency of physical exercise, % (stable, n = 248) (worsened, n = 340) Most recent dental visit, % (stable, n = 243) (worsened,n = 332) Toothbrushing frequency, % (stable, (worsened, n = 252) n = 343) Alcohol consumption, % (stable, (stable, % consumption, Alcohol Education, % (stable, n = 243) (worsened, n = (a) (worsened, n = 252) (stable, % Gender, n = (a) Basic 3 (a) TABLE characteristics of the study population [proportions/means and their standard deviations (SD)] categorized according to the fasting46 years plasma (b) Basic glucose characteristics levels and at 31 of the study population [proportions/means and their standard deviations (SD)] categorized by the level of education and31 46 years and the fasting plasma glucose levels at TEGELBERG et al. | 31 (Continues) n = 56) 0.9 (1.4) 53.8 1.4 (3.1) 1.5 (0.4) 28.7 (17.2) 5.3 (1.2) 5.6 (5.2) 16.4 50.9 32.7 1.3 (0.7) 26.7 (1.5) <5.0 mmol/l at and 31 5.6–6.9 mmol/l (prediabetes) at 46 years ( n = 179) 0.9 (1.3) 43.5 1.8 (5.9) 1.6 (0.4) 24.3 (16.5) 5.3 (0.9) 5.4 (6.3) 45 39.4 15.6 1.1 (0.5) 27.1 (1.4) Worsened glucose balance glucose Worsened <5.0 mmol/l at and 31 5.0– 5.59 mmol/l at 46 years ( = 24) 0.8 (1.5) 83.3 1.5 (2.7) 1.5 1.4 (0.5) 24.9 (18.2) 5.2 (0.9) 6.9 (7.7) 33.3 45.8 20.8 1.4 (0.8) 26.8 (1.5) 5.6–6.9 mmol/l (prediabetes) at and 31 46 years (n = 119) 0.9 (1.4) 65.6 2.2 (5.0) 1.6 (0.4) 27.1 (18.2) 5.3 (0.9) 6.4 (6.9) 47.9 34.5 17.6 1.1 (0.8) 26.9 (1.5) Stable glucose balance glucose Stable 5.0–5.59 mmol/l at and31 46 years (n 344) n = 109) 0.9 (1.4) 1.0 (2.3) 1.7 (0.3) 24.2 (16.1) 5.3 (0.8) 5.0 (5.8) 5.0 68.8 25.7 5.5 0.9 (0.5) 0.9 27.0 (1.5) Reference Reference category <5.0 mmol/l at 31 and 46 years ( 341) 344) 17.4 250) (worsened, (worsened, 250) ), % (stable, (worsened, n = 252) n = 2 344) 344) 344) % (stable, n = 67) (worsened, n = 82) mean (SD) mean (SD) (stable, (worsened, n = 252) n = (SD) (stable, (worsened, n = 252) n = 251) (worsened,n = 251) n = (SD) (stable, n = n = 343) 252) (worsened, n = 252) (stable, n = <24.9 25–29.9 ≥30 (stable, n = 227)(stable, (worsened, n = 302) (b) education Low Male Gender, 249) (worsened, n = 249) (stable, n = 340) Number of sites BL mm, ≥5 mean (SD) Number of sites with PPD mm, ≥4 249) (worsened, n = 249) (stable, n = HDL cholesterol (mmol/l), mean BOP, % of sitesBOP, (SD) Total cholesterol (mmol/l), mean Number of teeth, mean (SD) (stable, Number of teeth with plaque, mean Triglycerides (mmol/l), mean (SD) (SD) mean (mmol/l), Triglycerides BMI (kg/m Table 3 (Continued)Table 32 | TEGELBERG et al. mmol/l

(Continues) years

mmol/l at and 31 5.6–6.9

56)

n prediabetes) at 46 ( 30.8 55 <5.0 ( 61.5 25 4.2 (4.0) years

mmol/l at 46 mmol/l at and 31 5.0–

179)

.59 n 5 19.6 30.3 Worsened glucose balance glucose Worsened <5.0 ( 56.5 13.6 6.2 (9.9) years

mmol/l (prediabetes)

24)

t 31 andt 31 46 a 5.6–6.9 25 75 (n 50 18.2 11.8 (20.6) years

mmol/l at

119)

1 and 46 Stable glucose balance glucose Stable 5.0–5.59 3 21.9 43.2 (n 56.3 30 6.3 (7.6) years

mmol/l at 31

109)

nd 46 n a Reference Reference category <5.0 ( 4.3 13.3 73.9 4.5 3.6 (5.1) ) ≥30, 2 67) 82) 176) 255) 67) 82) 63) 78) 67) 80) years ago,

======

% (stable, n (worsened, n (worsened, n % (stable, n % (stable, n twice a day or more, % (stable, n (worsened, (worsened, n (worsened, n (worsened, n dental visit, visit, dental ≥3 consumption g/ mean day, (SD) (stable, n High education Male Gender, Toothbrushing Toothbrushing Most recent Alcohol BMI (kg/m Table 3 (Continued)Table TEGELBERG et al. | 33 mmol/l

years

mmol/l at and 31 5.6–6.9

56)

n prediabetes) at 46 ( 32.5 <5.0 ( 65 7.5 11.7 (18.6) years

mmol/l at 46 mmol/l at and 31 5.0–

179)

.59 n 5 14.4 Worsened glucose balance glucose Worsened <5.0 ( 85.5 10.2 5.6 (9.8) years

mmol/l (prediabetes)

24)

t 31 andt 31 46 a 5.6–6.9 16.7 (n 75 18.2 6.4 (5.4) years

mmol/l at

119)

1 and 46 Stable glucose balance glucose Stable 5.0–5.59 3 14.8 (n 71.6 15.2 5.4 (8.3) years

mmol/l at 31

109)

nd 46 n a Reference Reference category <5.0 ( 6 90.4 8.6 3.4 (5.5) ) ≥30, 176) 176)

2 =

176) 176) 255) 254) 171) 248) 170) 247) years ago,

======

% (stable, n (worsened, n % (stable, n twice a day or more % (worsened, (worsened, n dental visit, visit, dental ≥3 (worsened, (worsened, n (worsened, n consumption, g/ mean day, (SD) (stable, n Toothbrushing Toothbrushing n (stable, Most recent Alcohol Alcohol BMI (kg/m Abbreviations: FPG, fasting plasma glucose; BMI, body mass index; HDL, high-density lipoprotein; bleeding BOP, on probing. Table 3 (Continued)Table 34 | TEGELBERG et al.

TABLE 4 Associations of stable and worsened glucose balance with the numbers of sites with ≥4 mm probing pocket depths (PPD ≥4 mm) and alveolar bone loss (BL ≥5 mm) using Poisson regression models [relative risk (RR) with 95% confidence intervals (95% CI)] in the total population and in individuals with a high level of education. Only individuals with no missing data were included

Total population High education

Unadjusted RR Adjusted RRa Adjusted RRb

(95% CI) (95% CI) (95% CI)

n = 647 n = 647 n = 448

Stable glucose balance The two-class FPG categorization (ADA): PPD ≥4 mm < 5.6 mmol/l at 31 and 46 years (ref.) 1 1 1 5.6–6.9 mmol/l (prediabetes) at 31 and 46 years 1.1 (0.8–1.5) 0.9 (0.6–1.2) 0.3 (0.1–0.6) BL ≥5 mm < 5.6 mmol/l at 31 and 46 years (ref.) 1 1 1 5.6–6.9 mmol/l (prediabetes) at 31 and 46 years 1.2 (0.7–1.9) 1.1 (0.7–1.7) 1.0 (0.5–1.8) The three-class FPG categorization: PPD ≥4 mm < 5.0 mmol/l at 31 and 46 years (ref.) 1 1 1 5.0–5.59 mmol/l at 31 and 46 years 2.5 (2.0–3.1) 1.7 (1.3–2.2) 3.3 (2.4–4.5) 5.6–6.9 mmol/l (prediabetes) at 31 and 46 years 1.9 (1.3–2.8) 1.1 (0.7–1.6) 0.7 (0.3–1.5) BL ≥5 mm < 5.0 mmol/l at 31 and 46 years (ref.) 1 1 1 5.0–5.59 mmol/l at 31 and 46 years 1.0 (0.8–1.3) 0.9 (0.6–1.3) 1.1 (0.7–1.6) 5.6–6.9 mmol/l (prediabetes) at 31 and 46 years 1.2 (0.7–2.0) 0.9 (0.5–1.6) 1.0 (0.5–2.1) Worsened glucose balance The two-class FPG categorization (ADA): PPD ≥4 mm < 5.6 mmol/l at 31 and 46 years (ref.) 1 1 1 < 5.6 mmol/l at 31 and 5.6 – 6.9 mmol/l (prediabetes) at 1.0 (0.9–1.2) 1.0 (0.8–1.1) 0.9 (0.7–1.1) 46 years BL ≥5 mm < 5.6 mmol/l at 31 and 46 years (ref.) 1 1 1 < 5.6 mmol/l at 31 and 5.6 – 6.9 mmol/l (prediabetes) at 1.2 (0.9–1.4) 1.3 (1.0–1.6) 0.9 (0.6–1.2) 46 years The three-class FPG categorization: PPD ≥4 mm < 5.0 mmol/l at 31 and 46 years (ref.) 1 1 1 < 5.0 mmol/l at 31 and 5.0–5.59 mmol/l at 46 years 1.8 (1.4–2.2) 1.8 (1.4–2.2) 1.3 (0.9–1.7) < 5.0 mmol/l at 31 and 5.6 – 6.9 mmol/l (prediabetes) at 1.9 (1.4–2.5) 2.8 (2.0–3.8) 0.8 (0.5–1.3) 46 years BL ≥5 mm < 5.0 mmol/l at 31 and 46 years (ref.) 1 1 1 < 5.0 mmol/l at 31 and 5.0–5.59 mmol/l at 46 years 1.0 (0.8–1.3) 1.1 (0.8–1.4) 1.2 (0.8–1.6) < 5.0 mmol/l at 31 and 5.6 – 6.9 mmol/l (prediabetes) at 1.5 (1.1–2.2) 1.8 (1.2–2.8) 1.4 (0.8–2.3) 46 years

aAdjusted for gender, BMI, education, number of teeth with plaque and number of tooth sites (offset). bAdjusted for gender, BMI, number of teeth with plaque and number of tooth sites (offset). TEGELBERG et al. | 35 three-class categorization) at ages 31 and 46 were used as the refer- risk estimates was not more than 0.2 when the variables were added ence category. Gender, BMI, education and plaque—all associated into the models one by one. with both the outcome and the explanatory variable (FPG) (Tables 1, 2 and 3a)—were used as covariates; plaque as a continuous and the others as categorized variables. The number of tooth sites was used 4 | DISCUSSION as an offset variable. The analyses were carried out among the total population and among highly educated individuals (graduated from In this study, an important finding was that individuals who devel- a university or a university of applied sciences). The statistical analy- oped prediabetes between the ages of 31 and 46 had an increased ses were done using the IBM SPSS Statistics for Windows, version likelihood of periodontal pocketing and alveolar bone loss. A novel finding was that development of early hyperglycaemia, in which 22 (IBM Corp. https://www.ibm.com/se-en/analytics/). the FPG levels were slightly elevated but still under the threshold for prediabetes, was associated with the number of sites with PPD ≥4 mm. Our overall interpretation of these results is that the hyper- 3 | RESULTS glycaemia-induced destructive changes in periodontal tissues may start already at FPG levels below the diagnostic threshold for diabe- The basic characteristics of the study population are presented in tes mellitus and even at levels below the threshold for prediabetes. Tables 1–3. As to the long-term glucose balance, the basic charac- The results are in line with an earlier longitudinal study, which teristics are presented in relation to the three-class categorization was made in a 30–64-year-old never-smoking non-diabetic popula- (Table 3a). tion with no ≥4 mm periodontal pockets at the baseline (Timonen There were no essential differences in the mean number of teeth et al., 2013). According to that study, impaired glucose metabolism, between the glucose balance categories (Table 3a). Of the 252 indi- measured as insulin resistance and altered beta cell function, was as- viduals who remained in the same glucose balance category at ages sociated with periodontal pocket formation during a 4-year period. 31 and 46, 109 individuals (43%) had FPG <5.0 mmol/l, 119 (47%) Another, 5-year study by Chiu et al. (2015) in a 35–44-year-old pop- had slightly elevated FPG (5.0–5.59 mmol/l) and 24 (10%) had predi- ulation showed that individuals with prediabetes according to the abetes (FPG 5.6–6.9 mmol/l). Out of the 344 individuals whose FPG ADA criteria (FPG 100–125 mg/dl corresponding to 5.6–6.9 mmol/l), was <5.0 mmol/l at age 31, 179 (52%) had slightly elevated FPG (5.0– but without previously diagnosed periodontal disease defined using 5.59 mmol/l) and 56 (16%) developed prediabetes by the age of 46. the Community Periodontal Index (CPI) <3, had a 25% higher risk for Overall, after adjusting for gender, BMI, education and plaque, developing deepened periodontal pockets (CPI ≥3) when compared and using the two-class (ADA) categorization of glucose balance, to individuals with normoglycaemia (FPG <100 mg/dl corresponding only weak if any associations were found between a stable long- to <5.6 mmol/l). However, only weak if any associations between term glucose balance and periodontal status (Table 4). prediabetes and periodontal status were found in the present study When the stable glucose balance was based on the three-class when the ADA criteria were used. categorization, slightly elevated FPG (5.0–5.59 mmol/l) at ages 31 The hypothesis of the biologically plausible mechanisms between and 46 was associated with the number of sites with PPD ≥4 mm hyperglycaemia and periodontal tissue destruction comes primarily (RR 1.7, 95% CI 1.3–2.2). The corresponding figure for individuals from studies that investigate the link between diabetes mellitus and belonging to the highly educated group was RR 3.3 (95% CI 2.4–4.5). periodontitis. These studies reported hyperglycaemia-associated Those individuals whose FPG worsened from <5.6 mmol/l at age microvascular changes, alterations in bone and connective tissue 31 to 5.6–6.9 at age 46, that is those who developed prediabetes metabolism, and aberrant neutrophil functions as potential mech- according to the two-class ADA categorization, were likely to have anisms (Mealey & Oates, 2006), while the prevailing hypothesis more sites with BL ≥5 mm (RR 1.3, 95% CI 1.0–1.6) when compared emphasizes the role of a hyper-inflammatory state in the infected to those with FPG <5.0 mmol/l at ages 31 and 46 (reference group) periodontal tissues in diabetes (Polak & Shapira, 2018; Taylor et al., (Table 4). When using the three-class categorization, the RRs for PPD 2013). According to this hypothesis, the increased periodontal de- ≥4 mm were 1.8 (95% CI 1.4–2.2) and 2.8 (95% CI 2.0–3.8) in indi- struction during hyperglycaemia can be explained by an exaggerated viduals with a slightly elevated glucose level (FPG 5.0–5.59 mmol/l) expression of a variety of pro-inflammatory mediators as a result of and prediabetes (FPG 5.6–6.9 mmol/l), respectively, in the total pop- the accumulation of advanced glycosylation end products (AGEs) in ulation. The corresponding risk estimates (RRs) for BL ≥5 mm were periodontal tissues. In light of the present results, it appears that 1.1 (95% CI 0.8–1.4) and 1.8 (95% CI 1.2–2.8). In most cases, the risk these effects on the periodontium could in some cases be observed estimates (RRs) were lower among highly educated individuals when already at FPG levels below the threshold for prediabetes. compared to the risk estimates among the total population. Due to the complexity related to the shifts of individuals be- In order to study the effect of other potential confounders such tween various glucose balance categories over the course of time, as HDL, triglycerides and alcohol consumption on the strength of we preferred two practical approaches—studying individuals with the association, we made additional modelling. In these models, stable and worsened glucose balance (Table 4). Unlike the earlier made among the total population we observed that the change in longitudinal studies (Chiu et al., 2015; Timonen et al., 2013), in which 36 | TEGELBERG et al. the glucose balance was defined only at the baseline, the NFBC1966 lingual, as recommended by the current standards (Holtfreter et al. study data made it possible for us to determine the long-term glu- 2015), only four sites were measured in this study. It is likely that cose balance based on FPG levels at two points of time, ages 31 and by omitting measurements on the approximal palatal/lingual sites 46. Notwithstanding, the true exposure time by hyperglycaemia we underestimated the extent and severity of periodontal infection could not be determined precisely, especially among those whose (Susin et al. 2005). An obvious advantage in using continuous out- glucose balance worsened between the ages of 31 and 46. come variables in the regression analyses was that it reduced the In the case of a stable glucose balance, the risk for ≥4 mm probing effect of measurement error compared to a situation where the pocket depths was increased in individuals who had slightly elevated outcome is dichotomous. A shortcoming was that the overall repro- glucose levels, whereas the risk was not increased in those who had ducibility in the measurement of the outcome variables, numbers of prediabetes already at the age of 31 (Table 4). One explanation for sites with ≥4 mm probing pocket depth and ≥5 mm alveolar bone the latter finding could be the improved health behaviours between level, was on average only moderate. ages 31 and 46 for individuals who were informed of their adverse Finally, in this study the associations of FPG with periodontal findings within their bloodwork. Namely, individuals who had FPG status were generally stronger in individuals whose FPG level wors- >5.6 mmol/l (prediabetes), high cholesterol, or low haemoglobin at ened from ages 31 to 46 than in those who had elevated, although age 31 were informed of these findings and advised to contact their stable, FPG at both ages. It appeared that there were more obese health care centre or medical doctor for further clarifications on individuals in the worsened than in the stable group (Table 3a) and their health issues. the possible role of obesity in this context remains to be explored. It is well known that the level of education is strongly related to Within the limitations of this study, we conclude that impairment lifestyle and health behaviour (Clark & Royer, 2013; Cutler & Lleras- in glucose control in non-diabetic individuals is potentially associ- Muney, 2010). We also performed the analyses among individuals ated with periodontal pocketing and alveolar bone loss. Due to the with a high level of education. The group of individuals with a low uncertainty associated with the findings, future studies are needed level of education was unfortunately too small for statistical anal- to verify the association of early hyperglycaemia with periodontal yses. In the case of a stable glucose balance, highly educated indi- status. viduals whose FPG was slightly elevated had an exceptionally high risk for ≥4 mm probing pocket depths (RR 3.3). It is unclear whether ACKNOWLEDGEMENTS the observed relative risk represents the true value related to ele- The present study is part of the Northern Finland Birth Cohort 1966 vated FPG in this limited sample or whether it is a random finding study. The authors thank the late professor Paula Rantakallio for or a finding related to another, unexamined variable. It could be ex- launching NFBC1966, the participants of the NFBC1966, the NFBC pected that restriction to the highly educated group would produce project centre and Paula Pesonen for her contribution regarding the less biased risk estimates than those obtained in the total population dental examination data. because the study population is more homogenous. Obesity has been associated with periodontitis in earlier studies CONFLICT OF INTEREST (Chaffee & Weston, 2010). Obesity also contributes to the devel- The authors declare no potential conflict of interest with respect to opment of hyperglycaemia via insulin resistance and impaired pan- the authorship and/or publication of this article. creatic beta cell function (Grundy, 2016; Lauterbach & Wunderlich, 2017). In order to control for the confounding effect of obesity, we ORCID used BMI as a covariate in the regression models and carried out the Paula Tegelberg https://orcid.org/0000-0001-5259-1859 analyses among non-obese individuals (BMI <30) (data not shown). We also studied the confounding effect of HDL, triglycerides and REFERENCES alcohol consumption by adding them into the regression models. American Diabetes Association (2017). 2. Classification and diagnosis These complementary analyses showed that BMI, HDL, triglycerides of diabetes. Diabetes Care, 40, 11–24. https://doi.org/10.2337/ dc17-S005 and alcohol consumption had no essential effect on the results. In Baltzis, D., Eleftheriadou, I., & Veves, A. (2014). Pathogenesis and treat- addition, it is worth noting that the number of teeth was quite high ment of impaired wound healing in diabetes mellitus: New insights. and similar between the various categories of glucose balance. To Advances in Therapy, 31, 817–836. https://doi.org/10.1007/s1232 avoid biases due to smoking (Hujoel et al., 2002), all the analyses 5-014-0140-x Chaffee, B. W., & Weston, J. S. (2010). Association Between Chronic were conducted in a never-smoking population. We also took into Periodontal Disease and Obesity: A Systematic Review and Meta- account possible systemic inflammation by excluding individuals Analysis. Journal of Periodontology, 81(12), 1708–1724. https://doi. with systemic inflammatory diseases. org/10.1902/jop.2010.100321 Since the oral examination was included in the NFBC study pro- Chiu, S. Y., Lai, H., Yen, A. M., Frann, J. C., Chen, L. S., & Chen, H. H. tocol for the first time when the cohort was 46 years old, the data did (2015). Temporal sequence of the bidirectional relationship between hyperglycemia and periodontal disease: A commu- not permit us to determine the true incidence of pocket formation nity-based study of 5,885 Taiwanese aged 35–44 years. Acta and alveolar bone loss. Another disadvantage was that instead of Diabetologica, 52, 123–131. https://doi.org/10.1007/s0059 measuring six sites per tooth, that is three buccal and three palatal/ 2-014-0612-0 TEGELBERG et al. | 37

Clark, D., & Royer, H. (2013). The Effect of Education on Adult Mortality (2017). Cross-sectional associations of impaired glucose metabo- and Health: Evidence from Britain. The American Economic Review, lism measures with bleeding on probing and periodontitis. Journal 103, 2087–2120. https://doi.org/10.1257/aer.103.6.2087 of Clinical Periodontollogy, 44, 142–149. https://doi.org/10.1111/ Cutler, D. M., & Lleras-Muney, A. (2010). Understanding differences jcpe.12662 in health behaviors by education. Journal of Health Economics, 29, Polak, D., & Shapira, L. (2018). An update on the evidence for patho- 1–28. https://doi.org/10.1016/j.jhealeco.2009.10.003 genic mechanisms that may link periodontitis and diabetes. Journal Giacco, F., & Brownlee, M. (2010). Oxidative stress and diabetic complica- of Clinical Periodontology, 45, 150–166. https://doi.org/10.1111/ tions. Circulation Research, 107, 1058–1070. https://doi.org/10.1161/ jcpe.12803 CIRCRESAHA Susin, C., Kingman, A., & Albandar, J.M. (2005). Effect of partial record- Grundy, S. M. (2016). Metabolic Syndrome update. Trends in ing protocols on estimates of prevalence of periodontal disease. Cardiovascular Medicine, 26(4), 364–373. https://doi.org/10.1016/j. Journal of Periodontology, 76, 262–267. https://doi.org/10.1902/ tcm.2015.10.004 jop.2005.76.2.262. Holtfreter, B., Albandar, J. M., Dietrich, T., Dye, B. A., Eaton, K. A., Eke, Tammelin, T., Näyhä, S., Laitinen, J., Rintamäki, H., & Järvelin, M.-R. P. I., Papapanou, P. N., & Kocher, T. (2015). Standards for report- (2003). Physical activity and social status in adolescence as pre- ing chronic periodontitis prevalence and severity in epidemiologic dictors of physical inactivity in adulthood. Preventive Medicine, 37, studies – Proposed standards from the Joint EU/USA Periodontal 375–381. https://doi.org/10.1016/s0091-7435(03)00162-2 Epidemiology Working Group. Journal of Clinical Periodontology, 42, Taylor, J. J., Preshaw, P. M., & Lalla, E. (2013). A review of the evidence 407–412. https://doi.org/10.1111/jcpe.12392 for pathogenic mechanisms that may link periodontitis and dia- Hong, J. W., Noh, J. H., & Kim, D. J. (2016). The prevalence and associ- betes. Journal of Clinical Periodontology, 40, 113–134. https://doi. ated factors of periodontitis according to fasting plasma glucose org/10.1902/jop.2013.134005 in the Korean adults: The 2012–2013 Korea National Health and Timonen, P., Saxlin, T., Knuuttila, M., Suominen, A. L., Jula, A., Tervonen, Nutrition Examination Survey. Medicine, 95, e3226. https://doi. T., & Ylöstalo, P. (2013). Role of insulin sensitivity and beta cell func- org/10.1097/MD.0000000000 003226 tion in the development of periodontal disease in adults without Hujoel, P. P., Drangsholt, M., Spiekerman, C., & DeRouen, T. A. (2002). diabetes. Journal of Clinical Periodontology, 40, 1079–1086. https:// Periodontitis-systemic disease associations in the presence of doi.org/10.1111/jcpe.12162 smoking – Causal Or Coincidental? Periodontology, 2000(30), 51– University of Oulu (1966). Northern Finland Birth Cohort 1966. University 60. https://doi.org/10.1034/j.1600-0757.2002.03005.x of Oulu. http://urn.fi/urn:nbn:fi:att:bc1e5408-980e-4a62-b899- ISCED (1997). International Standard Classification of Education, United 43bec3755243. Nations Educational, Scientific and Cultural Organization. World Health Organization (2000). Obesity: Preventing and Managing King, G. L. (2008). The Role of Inflammatory Cytokines in Diabetes and the Global Epidemic. Report of a WHO Consultation. World Health its Complications. Journal of Periodontology, 79, 1527–1534. https:// Organization Technical Report Series, 894, 8–15. doi.org/10.1902/jop.2008.080246 World Health Organization (2013). Oral Healt Surveys: basic methods – Lamster, I. B., Cheng, B., Burkett, S., & Lalla, E. (2014). Periodontal find- 5th ed. 2013, Annex 7. ings in individuals with newly identified pre-diabetes or diabetes Yip, W. C. Y., Sequeira, I. R., Plank, L. D., & Poppitt, S. D. (2017). Prevalence mellitus. Journal of Clinical Periodontology, 41, 1055–1060. https:// of pre-diabetes across ethnicities: A review of impaired fasting glu- doi.org/10.1111/jcpe.12307 cose (IFG) and impaired glucose tolerance (IGT) for classification of Lauterbach, M. A. R., & Wunderlich, T. F. (2017). Macrophage function in dysglycaemia. Nutrients, 9, e1273. https://doi.org/10.3390/nu911 obesity-induced inflammation and insulin resistance. Pflugers Archiv, 1273 469(3), 385–396. https://doi.org/10.1007/s00424-017-1955-5 Löe, H. (1993). Periodontal Disease. The sixth complication of diabetes mellitus. Diabetes Care, 16, 329–334. How to cite this article: Tegelberg P, Tervonen T, Mealey, B. L., & Oates, T. W. (2006). Diabetes mellitus and periodontal Knuuttila M, et al. Association of hyperglycaemia with diseases. Journal of Periodontology, 77, 1289–12303. https://doi. org/10.1902/jop.2006.050459 periodontal status: Results of the Northern Finland Birth Nathan, D. M., Davidson, M. B., DeFronzo, R. A., Heine, R. J., Henry, R. Cohort 1966 study. J Clin Periodontol. 2021;48:25–37. https:// R., Pratley, R., & Zinman, B. (2007). Impaired fasting glucose and doi.org/10.1111/jcpe.13388 impaired glucose tolerance: implications for care. Diabetes Care, 30, 753–759. https://doi.org/10.2337/dc07-9920 Pérez, C. M., Muñoz, F., Andriankaja, O. M., Ritchie, C. S., Martínez, S., Vergara, J., Vivaldi, J., López, L., Campos, M., & Joshipura, K. J. Received: 24 April 2020 | Revised: 27 August 2020 | Accepted: 12 October 2020 DOI: 10.1111/jcpe.13391

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

The association between metabolic syndrome and periodontitis in Spain: Results from the WORALTH (Workers’ ORAL healTH) Study

1ETEP (Etiology and Therapy of Periodontal and Peri-implant Diseases) Abstract Research Group, University Complutense Introduction: Evidence of an association between periodontitis and MetS (metabolic of Madrid, Spain 2Ibermutua, Madrid, Spain syndrome) remains controversial. The objective of this study is to evaluate the asso- 3Cualtis, Madrid, Spain ciation between periodontitis and MetS in a cross-sectional population survey. Material and Methods: WORALTH (Workers’ ORAL healTH) Study is a cross-sec- Correspondence Eduardo Montero, Facultad de tional survey, conducted on a representative sample of the Spanish employed popu- Odontología, Universidad Complutense lation, including 5154 participants. An oral examination following the World Health de Madrid, Plaza de Ramón y Cajal s/n, 28040 Madrid, Spain. Organization (WHO) criteria evaluated the periodontal status using the Community Email: [email protected] Periodontal Index (CPI) and Clinical Attachment Levels (CAL). Logistic regression anal- Funding information ysis with adjustment for potential confounders was used to evaluate the association This study was supported by Cualtis, previously named Sociedad de Prevención between periodontitis and MetS, and its individual components. Ibermutuamur, a company that focuses Results: Participants presenting a CPI = 4 were more likely to have MetS than subjects specifically on preventing diseases and accidents, by monitoring workers’ health. with CPI < 4 [odds ratio, OR = 1.41; 95% confidence interval (CI) 1.10–1.81; p < 0.001]. High blood pressure was the component with stronger association with periodontal status (OR = 1.94 for CAL ≥6 mm; 95% CI 1.49–2.53; p < 0.001). After stratifying for sex, the association was higher in women (OR = 2.20 for CPI = 4; 95% CI 1.31–3.62; p < 0.001). Non-metabolically healthy subjects, obese or not, presented a worse periodontal condition. Conclusion: Severe periodontitis (CPI = 4) was associated with MetS in a representative sample of the Spanish employed population. This association seems to be independent of body mass index and other potential confounders.

KEYWORDS dental health surveys, high blood pressure, impaired glucose regulation, metabolic syndrome, obesity, periodontitis

38 | wileyonlinelibrary.com/journal/jcpe J Clin Periodontol. 2021;48:38–50. MONTERO ET AL. | 39

1 | INTRODUCTION Clinical Relevance Periodontitis is a chronic inflammatory disease of infectious origin, Scientific rationale for study: There is a need to study defined by destruction of the tooth supporting tissues, which if not whether metabolic syndrome (MetS) and periodontitis are treated, may lead to tooth loss. Results from the Global Burden of associated and whether this potential association just ap- Disease (GBD) Study, and other epidemiological data, have reported plies for certain subgroups depending on age, sex or pres- that more than 50% of adults are affected by mild-to-moderate ence of other comorbidities. forms of periodontitis, while the severe forms evidence a global Principal findings: A worse periodontal condition is associ- age-standardized prevalence of ≈11%, what results in the sixth most ated with significantly higher odds for suffering MetS, in- prevalent condition in the world (Kassebaum et al., 2014; Petersen dependently of the syndrome definition, being high blood & Ogawa, 2012). pressure the component with a stronger association with Non-communicable diseases (NCDs) like cardiovascular dis- periodontitis. ease (CVD), diabetes mellitus (DM), cancer or chronic respiratory Practical implications: Periodontitis and MetS are signifi- disease are currently the most prevalent diseases affecting hu- cantly associated. Therefore, regular periodontal check- mans, what is leading to a health and socio-economic global cri- ups should be done in patients with MetS. The potential sis (Beaglehole et al., 2011). Also, periodontitis is considered as benefits of periodontal therapy in these patients need to a NCD (Herrera et al., 2018; Tonetti et al., 2017) and different be determined. risk factors associated with periodontitis may be shared by the other NCDs, such as smoking, stress, unhealthy diet, glycaemic control, or genetic and socio-economic determinants (Petersen & Ogawa, 2012; Pihlstrom et al., 2005). In addition to the shared risk factors, periodontitis is associated with a low-grade systemic in- Moreover, different phenotypes of MetS have been described de- flammatory response that increases the risk for atherogenesis and pending on the relation between the body size and the metabolic hyperglycaemia (Morita et al., 2010; Paraskevas et al., 2008; Reyes profile (Karelis et al., 2004). “Metabolically healthy obese” (MHO) et al., 2013). Different specific mechanisms and pathological path- are subjects that, in spite of having obesity, lack any cardiometabolic ways have been attributed for this significant association between abnormality. Also, the opposite [“Metabolically non-healthy non- periodontitis and CVD or DM, either through direct mechanisms obese” (MNHNO)] refers to those subjects metabolically abnormal, (bacteraemia or endotoxaemia, e.g. by periodontal pathogens such but with normal weight. as Porphyromonas gingivalis), and/or indirectly, through systemic The potential association between periodontitis and MetS has inflammation or molecular mimicry-induced autoimmune damage been recently reviewed in a systematic review with meta-analysis, (de Pablo et al., 2009; Reyes et al., 2013; Schenkein & Loos, 2013; reporting an odds ratio (OR) of 1.71 [95% confidence interval (CI) Taylor et al., 2013). 1.42–2.03] (Nibali et al., 2013). Additionally, conditions associated Metabolic syndrome (MetS) consists of a cluster of clinical and with MetS, such as erectile dysfunction (Gorgel et al., 2014), have biological abnormalities, including insulin resistance, central obe- also been proposed to be associated with periodontitis (Bizzarro & sity, dyslipidaemia and hypertension that constitute a risk factor for Loos, 2019; Zadik et al., 2009). However, still the association be- CVD (Eckel et al., 2005). In fact, subjects affected with MetS have tween MetS and periodontitis remains controversial, since the diag- twice the risk of developing CVDs (Benguigui et al., 2010). Several nostic criteria used to define the case of MetS and of periodontitis studies have reported an increase in MetS prevalence around the vary between studies, and some studies have only been able to find world, with a particularly important rise in developing countries association in specific groups, such as women, diabetic patients (Cameron et al., 2004; Misra & Khurana, 2008; Prasad et al., 2012). or in specific age groups (D'Aiuto et al., 2008; Furuta et al., 2013; In Spain, the most recent population study reported a prevalence Sora et al., 2013; Tu et al., 2013), while others have observed no of 31.2% in adults aged 35–74 years (Martinez-Larrad et al., 2016), association between MetS and periodontitis (Benguigui et al., 2010; which is in a similar range compared with recent European estimates LaMonte et al., 2014; Zuk et al., 2017). Hence, there is a need of (Han & Lean, 2016). Recently, the International Diabetes Federation further research to clarify these inconsistencies and heterogeneous (IDF) has re-defined the MetS, addressing both clinical and research results among distinct groups. needs, since the existence of multiple definitions had proven to It was, therefore, the objective of the present epidemiological be confusing and to hinder direct comparisons between studies investigation to evaluate a representative sample of the Spanish (International Diabetes Federation, 2015). According to the new IDF employed population to study the association between their peri- definition, to be diagnosed with MetS, a patient must present central odontal status and the presence of MetS. As secondary objective, obesity (defined as waist circumference, with specific values accord- we also evaluated the prevalence of periodontitis among MHO ing to ethnicity) plus any two of the following three factors: (a) lipid and MNHNO subjects. This study was part of a wider survey, the abnormality, (b) high blood pressure (BP) and (c) hyperglycaemia. WORALTH (Workers’ ORAL healTH), studying the oral health 40 | MONTERO et al. status and oral healthcare needs of the Spanish adult employed 2.3 | Physical examination population (Carasol et al., 2016). Waist circumference (WC) was measured while the subject was standing, using the midpoint between the lowest rib and the iliac 2 | METHODS crest as a reference. Additional data from physical examinations included weight, height and two measurements of BP. Body mass 2.1 | Study design index (BMI) was calculated, and subjects were stratified in three groups according to its value: obese (≥30 kg/m2), overweight (25– The WORALTH Study was conducted on a representative sample of 29 kg/m2) and normal (<25 kg/m2). the employed population in Spain, from April 2008 to June 2011. It is an epidemiological survey using the WHO criteria for Oral Health Surveys (WHO, 1997), with a specific methodology to assess the sub- 2.4 | Biochemical analysis ject's periodontal condition, which has been reported in a previous publication (Carasol et al., 2016). In brief, workers, within the context Serum samples were analysed in reference laboratories using stand- of a broader epidemiological study on cardiovascular risk assessment, ard protocols, variation coefficients and ranges accepted by the were examined during their regular annual health evaluation by means Spanish Society of Clinical Biochemistry and Molecular Pathology's of a structured interview, physical examination and laboratory deter- (SEQC) (Sacks et al., 2011). Subjects were advised to fast 12 hours minations (ICARIA, Ibermutuamur CArdiovascular RIsk Assessment) prior to taking the blood sample. The biochemical determinations in- (Sanchez-Chaparro et al., 2006, 2008, 2011; Valdivielso et al., 2009). cluded fasting plasma glucose (FPG), triglycerides, total cholesterol After applying a proportionate stratified random sampling method, and high-density lipoprotein (HDL) cholesterol. Low density lipopro- depending on geographical area, age and gender, 5201 workers were tein (LDL) cholesterol was also calculated using Friedewald's equa- screened for inclusion and, finally, 5130 subjects were included in the tion (Friedewald et al., 1972). oral health examination, since 47 subjects refused to participate and 24 were fully edentulous. The sample size of each segment had been previously calculated in relation to the Spanish Labour Force Survey, 2.5 | Oral and periodontal examination 2nd quarter (Instituto Nacional de Estadística, 2008). The protocol of this investigation was reviewed and approved by Ibermutuamur Following an oral examination using the WHO criteria (WHO, 1997), Ethics Committee. This study conforms with the Strengthening the the periodontal condition was assessed by means of the Community Reporting of Observational Studies in Epidemiology (STROBE) guide- Periodontal Index (CPI) and the registration of clinical attachment lines for reporting cross-sectional studies. levels (CAL). The ten index teeth were assessed at three buccal sites (mesiobuccal, midbuccal, distobuccal) and three lingual sites (mesio- lingual, midlingual, distolingual) and the highest value was recorded 2.2 | Socio-demographic and behavioural variables at each sextant. Training and calibration sessions were conducted by an experienced WHO epidemiologist, and crossed-examinations These variables were obtained from the medical examination (age, with gold standard were carried out to determine the degree of in- gender, smoking status) and the oral health questionnaire (country ter-examiner agreement. Calibration data were provided in a previ- of origin, education and income levels, dental visits). Subjects were ous publication (Carasol et al., 2016). stratified in different categories for each variable, as follows:

• Age: <25 years old, 25–34 years, 35–44 years, 45–54 years, or 2.6 | Metabolic syndrome definitions ≥55 years. • Smoking habits: never, former (who quit at least 12 months be- Two different MetS definitions were used in this study: i) the one fore) or current smokers. released by the National Cholesterol Education Program Adult • Country of origin: Spain or other countries. Treatment Panel III (NCEP-ATP III) (National Cholesterol Education • Occupation: white-collar (non-manual occupations) or blue-collar Program Expert Panel on Detection & Treatment of High Blood (manual occupations). Cholesterol in, 2002); and ii) the standardized diagnostic crite- • Education level: low (primary school), medium (secondary school) ria for MetS according to the IDF consensus worldwide definition or high (university). (International Diabetes Federation, 2015). • Income level (based on the net income of the family unit): ≤1200 According to the NCEP-ATP III (National Cholesterol Education €/month, 1.200–3.600 €/month and >3.600 €/month. Program Expert Panel on Detection & Treatment of High Blood • Visits to the dentist: regular visits (at least once during the previ- Cholesterol in, 2002), a subject is diagnosed of MetS if he/she has ous year) or irregular attendance. three or more of the following criteria: MONTERO ET AL. | 41

• Central obesity (defined as waist circumference ≥102 cm in men two different models were calculated: (a) a crude model and (b) an and ≥88 cm in women). adjusted model considering all potential confounders. The analy- • Hypertriglyceridaemia (≥150 mg/dl). ses were carried out using STATA version 13.1 with SVY package • Low HDL cholesterol (<40 mg/dl in males, <50 mg/dl in females). (StataCorp). • High BP (systolic BP ≥130 mmHg or diastolic BP ≥85 mmHg). • High fasting glucose (>110 mg/dl). 3 | RESULTS According to the IDF consensus definition (International Diabetes Federation, 2015), a subject with MetS is defined when The sample consisted of 4353 subjects, with 2623 men (60.1%) and presenting central obesity (defined as waist circumference ≥94 cm 1730 women (39.9%). Demographic and clinical characteristics of in males and ≥80 cm in females, in populations of European origin), the sample by MetS status are shown in Table 1. Overall, prevalence and any two of the following factors: of MetS was 8.5% using NCEP-ATP III criteria and 16.2% using IDF criteria. Subjects with MetS were older, had lower educational level • High triglyceride levels (≥150 mg/dl or specific treatment for this and were smokers or former smokers, when compared to subjects lipid abnormality). without MetS. In addition, patients with MetS (with both definitions) • Reduced HDL cholesterol (<40 mg/dl in males, <50 mg/dl in fe- presented higher BP, more abdominal obesity, lower levels of HDL males or specific treatment for this lipid abnormality). and higher levels of triglycerides and fasting plasma glucose, which • High BP (systolic BP ≥130 mmHg, or diastolic BP ≥85 mmHg or are components of the syndrome definition. A limited proportion of treatment of previously diagnosed hypertension). patients were undergoing drug treatment for diabetes (n = 87; 2.0%), • High fasting plasma glucose (fasting plasma glucose ≥100 mg/dl high blood pressure (n = 221; 5.1%) or lipid-lowering drugs (n = 288; or previously diagnosed of type 2 diabetes). 6.6%). Periodontal evaluation and oral health behaviour indicators are shown in Table 2. The percentage of subjects with periodontal pock- 2.7 | Body size phenotype definitions ets (CPI codes 3–4) was significantly higher among patients with MetS, when compared with patients without MetS, independently In the present study, we used the criteria proposed by Wildman of the definition (52.0% versus 35.1%, respectively, with IDF criteria, et al. (Wildman et al., 2008) for defining body size phenotypes. p < 0.001; 52.5% versus 36.4%, respectively, with ATP III criteria, Metabolically abnormal phenotype was defined by the presence p < 0.001). The same occurred for the proportion of subjects with of ≥2 cardiometabolic abnormalities (from those ones included in attachment loss ≥6 mm (12.9% versus 6.0%, respectively, with IDF the IDF consensus definition for MetS). The study population was criteria, p < 0.001; 14.4% versus 6.5%, respectively, with ATP III cri- stratified by BMI categories (normal weight and overweight versus teria, p < 0.001), or with the number of remaining teeth (subjects obese) and metabolic phenotype, thus generating four body size diagnosed of MetS presented significantly lower number of teeth; phenotypes: metabolically healthy non-obese (MHNO), metaboli- p < 0.001). Subjects brushing their teeth at least twice a day were cally healthy obese (MHO), metabolically non-healthy non-obese less frequently diagnosed of MetS, independently of the definition, (MNHNO) and metabolically non-healthy obese (MNHO). when compared with subjects with MetS (46.0% versus 54.0%, using the IDF criteria, p < 0.001; or 45.1% versus 54.9%, respectively, using the ATP III criteria, p < 0.001). 2.8 | Data analysis Table 3 and S1 describe the characteristics of the participants for each MetS component (based on the IDF definition) according to the Data from subjects without enough information for the diagnosis of CPI or CAL category. Subjects who have as highest code a CPI = 3–4 MetS were excluded from the analyses (n = 777). Hence, the final were significantly associated with more abdominal obesity, hyper- analyses included 4353 participants. Descriptive statistics were triglyceridaemia, high BP and low HDL cholesterol (p < 0.001). CPI calculated for demographic and clinical characteristics, by diagno- explained 4.3% of the variability in waist circumference and 5.2% sis of MetS, according to the different definitions, using chi-squared of the variability in systolic BP. Similar findings were observed for tests, for categorical variables, and t tests and analysis of variance, subjects with CAL ≥6 mm. In general, subjects with CPI codes 3–4 for continuous variables. Similar analyses were performed for the or CAL ≥4–5 mm were associated with more positive components different body size phenotypes. Different sets of binomial logistic among those included in the definition of MetS. regression analyses were carried out using the following dependent The association between the different MetS definitions, their in- variables: (a) subjects who had as highest code CPI = 4, (b) subjects dividual components and severe periodontitis (defined as CPI = 4 or with CAL ≥ 6 mm and (c) subjects with ≤15 missing teeth versus sub- CAL ≥6 mm) are presented in Tables 4 and 5. After adjusting for sex, jects with ≥16 missing teeth. The independent variables were either age, smoking habit, educational level and toothbrushing frequency, the different definitions of MetS, their components or the body in those subjects with CPI = 4 (Table 4), the prevalence of MetS was size phenotype. For each of the sets of logistic regression analyses, higher (OR = 1.41; 95% CI 1.10–1.81; p = 0.007, with IDF definition; 42 | MONTERO et al.

TABLE 1 Demographic and clinical characteristics of study participants by Metabolic Syndrome (MetS) diagnosis

NCEP-ATP III Criteria IDF Criteria

Characteristic All Without MetS With MetS p Without Met With MetS p

Number of 4353 (100%) 3985 (91.6%) 368 (8.5%) 3646 (83.8%) 707 (16.2%) participants Age group (years) <25 390 (9.0%) 379 (9.5%) 11 (3.0%) <0.001 376 (10.3%) 14 (2.0%) <0.001 25–34 1374 (31.6%) 1316 (33.0%) 58 (15.8%) 1279 (35.1%) 95 (13.4%) 35–44 1289 (29.6%) 1191 (29.9%) 98 (26.6%) 1101 (30.2%) 188 (26.6%) 45–54 877 (20.2%) 765 (19.2%) 112 (30.4%) 639 (17.5%) 238 (33.7%) ≥55 423 (9.7%) 334 (8.4%) 89 (24.2%) 251 (6.9%) 172 (24.3%) Male sex, % 2623 (60.1%) 2345 (58.9%) 278 (75.6%) <0.001 2091 (57.4%) 532 (75.3%) <0.001 Smoking habit, % Non smoker 2081 (51.1%) 1939 (52.0%) 142 (41.3%) <0.001 1815 (53.3%) 266 (39.7%) <0.001 Former smoker 608 (14.9%) 534 (14.3%) 74 (21.5%) 454 (13.3%) 154 (23.0%) Smoker ≤10 cig/ 686 (16.8%) 634 (17.0%) 52 (15.1%) 578 (17.0%) 108 (16.1%) day Smoker >10 cig/ 701 (17.2%) 625 (16.8%) 76 (22.1%) 559 (16.4%) 142 (21.2%) day Occupation, % White-collar 2297 (52.8%) 2123 (53.2%) 180 (47.6%) 0.036 1947 (53.4%) 350 (49.5%) 0.058 Blue-collar 2056 (47.2%) 1871 (46.9%) 198 (52.4%) 1699 (46.6%) 357 (50.5%) Country of origin, % Spain 3752 (89.4%) 3430 (89.2%) 322 (91.5%) 0.185 3121 (88.6%) 631 (93.3%) <0.001 Other countries 445 (10.6%) 415 (10.8%) 30 (8.5%) 400 (11.4%) 45 (6.7%) Education, % Primary School 1044 (24.9%) 916 (23.8%) 128 (36.8%) <0.001 792 (22.5%) 252 (37.6%) <0.001 Secondary 1765 (42.1%) 1625 (42.2%) 140 (40.2%) 1509 (42.8%) 256 (38.2%) School University 1388 (33.1%) 1308 (34.0%) 80 (22.3%) 1225 (34.7%) 163 (24.3%) Net income (monthly), % <1200 € 1201 (31.1%) 1108 (31.3%) 93 (29.6%) 0.419 1026 (31.6%) 175 (28.7%) 0.082 1201–3600 € 2210 (57.3%) 2020 (57.0%) 190 (60.5%) 1836 (56.5%) 374 (61.3%) >3600€ 447 (11.6%) 416 (11.7%) 31(9.9%) 386 (11.9%) 61 (10.0%) BMI (kg/m2) Normal 1979 (45.5%) 1944 (48.8%) 35 (9.5%) <0.001 1930 (52.9%) 49 (6.9%) <0.001 Overweight 1657 (38.1%) 1520 (38.1%) 137 (37.2%) 1318 (36.2%) 339 (48.0%) Obese 717 (16.5%) 521 (13.1%) 196 (53.3%) 398 (10.9%) 319 (45.1%) Waist 87.7 ± 12.9 86.3 ± 12.0 102.7 ± 12.2 <0.001 85.1 ± 11.7 101.0 ± 10.1 <0.001 circumference (cm) Triglycerides (mg/ 105.0 ± 80.2 97.1 ± 65.6 219.1 ± 151.7 <0.001 92.7 ± 62.7 176.3 ± 122.1 <0.001 dl) HDL (mg/dl) 59.1 ± 14.7 59.9 ± 14.6 47.8 ± 11.6 <0.001 60.2 ± 14.7 52.7 ± 13.1 <0.001 Systolic/diastolic 121 ± 17/75 ± 11 119 ± 16/74 ± 11 138 ± 16/86 ± 11 <0.001 118 ± 15/74 ± 10 134 ± 17/84 ± 11 <0.001 blood pressure (mm Hg) Fasting plasma 86.6 ± 15.3 85.3 ± 12.5 105.4 ± 31.0 <0.001 84.5 ± 12.3 98.8 ± 23.2 <0.001 glucose (mg/dl)

Abbreviations: BMI, body mass index; cig, cigarette; HDL, high-density lipoprotein; IDF, International Diabetes Federation; NCEP-ATP, National Cholesterol Education Program Adult Treatment Panel III. MONTERO ET AL. | 43

TABLE 2 Periodontal measures and oral health behaviours of study participants by Metabolic Syndrome (MetS) diagnosis

NCEP-ATP III Criteria IDF Criteria

Characteristic All Without MetS With MetS p Without MetS With MetS p

CPI: Percentage of subjects who have as highest code (%) Code 0 178 (4.1%) 167 (4.2%) 11 (3.0%) <0.001 165 (4.5%) 13 (1.8%) <0.001 Code 1 343 (7.9%) 330 (8.3%) 13 (3.5%) 320 (8.9%) 23 (3.3%) Code 2 2187 (50.2%) 2036 (51.1%) 151 (41.0%) 1883 (51.7%) 304 (43.0%) Code 3 1224 (28.1%) 1100 (27.6%) 124 (33.7%) 979 (26.9%) 245 (34.7%) Code 4 421 (9.7%) 352 (8.8%) 69 (18.8%) 299 (8.2%) 122 (17.3%) Percentage of subjects with CAL (%) 0–3 mm 3455 (79.37%) 3211 (80.6%) 244 (66.3%) <0.001 2988 (82.0%) 467 (66.1%) <0.001 4–5 mm 587 (13.5%) 516 (13.0%) 71 (19.3%) 438 (12.0%) 129 (21.1%) ≥6 mm 311 (7.1%) 258 (6.5%) 53 (14.4%) 220 (6.0%) 91 (12.9%) Number of teeth 25.6 ± 3.8 25.8 ± 3.5 23.8 ± 5.6 <0.001 25.9 ± 3.4 24.0 ± 5.0 <0.001 Dental visits

Abbreviations: NCEP-ATP, National Cholesterol Education Program Adult Treatment Panel III; IDF, International Diabetes Federation; CPI, Community Periodontal Index; CAL, clinical attachment level.

and OR=1.58; 95% CI 1.16–2.16; p = 0.004, with ATP III criteria). status in males, with the exception of subjects with CPI=4, and ac- When severe periodontitis was defined as CAL ≥6 mm (Table 5), sim- cording to the NCEP-ATP III definition (OR = 1.44; 95% CI 1.01–2.04; ilar results were observed (OR = 1.37; 95% CI 1.04–1.82; p = 0.027, p = 0.043). with IDF definition; and OR = 1.57; 95% CI 1.11–2.21; p = 0.010, with Table S2 depicts the prevalence of the different metabolic-obe- ATP III criteria). Among the individual components, only high BP was sity phenotypes in the studied population. Out of the 4353 subjects, associated with severe periodontitis (OR = 1.44; 95% CI 1.14–1.81; 66.1% (n = 2879) were MHNO, 17.4% (n = 757) were MNHNO and p = 0.002, for CPI = 4; and OR = 1.94; 95% CI 1.49–2.53; p ≥ 0.001, among those obese there was a higher percentage of MHO (9.1%; for CAL ≥ 6 mm), although statistically significant associations were n = 394)) than MNHO (7.4%; n = 323). Subjects without metabolic observed between the number of positive MetS components (two health, either obese or not, were more likely to be older, smokers or more) and severe periodontitis (OR=1.41; 95% CI 1.12–1.77; and to present a worse periodontal condition. The percentage of p = 0.003, for CPI = 4; and OR = 1.47; 95% CI 1.13–1.90; p = 0.004, subjects with deep periodontal pockets (6 mm, CPI code 4)) was sig- for CAL ≥6 mm). The analysis of the relationship of MetS with tooth nificantly higher (p ≥ 0.001) in the metabolically unhealthy groups loss showed that MetS, as determined by the IDF criteria, was as- (14.0% for MNHNO and 19.5% for MNHO) when compared to the sociated with tooth loss, with an OR of 1.89 (95% CI 1.05–3.40; metabolically healthy ones (7.4% for MHNO and 10.2% for MHO), p = 0.033), after adjusting for confounders. independently of the body mass index category. Similar findings After stratifying by sex and adjusting for potential confounders, were observed when CAL was used as the periodontal criterion. females showed greater ORs of having MetS, independently of the The association between periodontal measures (CPI or CAL) and definition (Tables 4 and 5). For the IDF criteria, females with CPI = 4 body size phenotypes is presented in Table 6. After adjusting for presented an OR = 2.20 (95% CI 1.31–3.62; p = 0.003) of suffer- confounders, the results of the multinomial logistic regression anal- ing MetS. No association was found between MetS and periodontal ysis showed that severe periodontitis was significantly associated 44 | MONTERO et al. 2 0.029 0.007 Adjusted R 0.043 0.014 0.033 0.024 0.052/0.035 <.001 <.001 0.152 0.375 <0.001 p value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 * * * * * * * * * 68.4 ± 14.2 134.9 ± 126.0 54.3 ± 12.4 ± 14.2 57.6 Code 4 (%) 92.6 ± 12.0 95.3 ± 10.8 84.3 ± 11.8 92.4 ± 25.4 89 (22.0%) 240 (57.0%) (37.6%) 152 (52.9%) 217 (22.7%)91 127 (30.2%) 125 (29.7%) 99 (23.5%) 54 (12.8%) (3.8%)16 129 ± 18/79 ± 11 * * * * * * * * * 65.6 ± 15.1 115.3 ± 88.0 54.0 ± 12.2 ± 14.3 57.9 220 (18.7%) 489 (41.1%) 473 (38.6%)473 Code 3 (%) 390 (31.9%) 107 (8.7%)107 90.0 ± 12.3 ± 10.894.1 82.2 ± 10.9 ± 15.3 87.4 629 (51.4%) (31.4%)371 31 (2.5%)31 201 (17.1%) 223 (18.2%) ± 11 ± 17/77 124 ± 15.3 67.2 98.5 ± 67.1 53.6 ± 11.6 ± 14.9 59.3 289 (13.7%) 606 (28.7%) 1100 (50.3%) Code 2 (%) 607 (27.8%) 115 (5.3%)115 92.1 ± 11.4 ± 10.9 79.1 86.6 ± 12.9 86.0 ± 13.3 907 (41.2%) 490 (23.4%) 26 (1.2%)26 339 (15.5%) 349 (16.8%) ± 11 ± 16/74 119 p < 0.01). 68.5 ± 13.7 82.1 ± 42.5 54.1 ± 13.1 62.4 ± 15.2 38 (11.8%) 69 (20.9%)69 220 (64.1%) Code 1 (%) 82 (23.9%) 6 (1.8%) ± 11.4 89.9 ± 8.7 76.7 82.2 ± 11.9 81.8 ± 9.2 114 (33.2%)114 44 (13.6%) 2 (0.6%) 33 (9.6%) 23 (7.1%) ± 10 ± 16/71 115 , standard deviation; HDL, high-density lipoprotein.

a a a a a a

a a , %) n 104 (58.4%) ± 13.5 67.1 41 (24.1%) 41 ± 49.5 87.0 45 (25.3%)45 56.2 ± 12.1 ± 13.9 61.9 20 (11.2%) 24 (14.4%) 24 7 (3.9%) 2 (1.1%) Percentage of subjects who have as highest code Code 0 (%) ± 10.7 91.1 ± 12.077.3 (8.3%)14 71 (39.9%) 71 84.2 ± 13.3 83.7 ± 9.9 35 (20.6%) ± 15/73 ± 10 116 ) SD ) ; %) n ,%) ; %) n n ; %) n ; %) ) ) ) Characteristics of the participants for the Metabolic Syndrome (MetS) components (according to the IDF definition) by Community Periodontal Index (CPI) mean ± SD mean ± SD pressure (mm Hg; mean ± SD 0 Women Waist circumference (cm; Fasting blood glucose (mg/dl; 1 Men HDL (mg/dl; mean ± SD 2 Triglycerides (mg/dl; mean ± blood Systolic/diastolic 3 4 Men Women Number of positive MetS components ( High blood pressure ( Low HDL cholesterol ( Abdominal obesity ( Hypertriglyceridaemia( Hyperglycaemia( n Reference category. *Statistically significant difference when compared with the reference category ( TABLE 3 Abbreviations: International IDF, Diabetes Federation; SD a MONTERO ET AL. | 45

TABLE 4 Association between the different definitions of Metabolic Syndrome (MetS) and its individual components with severe periodontitis (defined as CPI code 4)

Total Male gender Female gender

p OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) Value

MetS (IDF definition) 1.41 (1.10–1.81) 0.007 1.25 (0.94–1.66) 0.128 2.20 (1.31–3.62) 0.003 Abdominal obesity 1.19 (0.95–1.49) 0.126 1.08 (0.83–1.40) 0.563 1.50 (0.96–2.34) 0.075 Hypertriglyceridaemia 1.17 (0.93–1.48) 0.188 1.09 (0.83–1.43) 0.523 1.45 (0.88–2.38) 0.148 High blood pressure 1.44 (1.14–1.81) 0.002 1.28 (0.99–1.66) 0.064 2.18 (1.39–3.41) 0.001 Hyperglycaemia 1.10 (0.84–1.44) 0.490 1.04 (0.77–1.42) 0.789 1.25 (0.70–2.25) 0.451 Low HDL cholesterol 1.40 (1.06–1.83) 0.015 1.60 (1.18–2.17) 0.003 0.92 (0.51–1.68) 0.788 ≤1 component vs ≥2 1.41 (1.12–1.77) 0.003 1.26 (0.95–1.69) 0.114 1.58 (1.03–2.41) 0.034 components MetS (NCEP-ATP III definition) 1.58 (1.16–2.16) 0.004 1.44 (1.01–2.04) 0.043 2.33 (1.24–4.39) 0.009 Abdominal obesity 1.19 (0.93–1.52) 0.174 1.16 (0.87–1.56) 0.316 1.22 (0.77–1.93) 0.388 Hypertriglyceridaemia 1.13 (0.89–1.45) 0.317 1.03 (0.78–1.36) 0.817 1.55 (0.91–2.66) 0.109 High blood pressure 1.43 (1.14–1.79) 0.002 1.26 (0.97–1.62) 0.080 2.27 (1.45–3.54) <.001 Hyperglycaemia 1.39 (1.00–1.94) 0.052 1.41 (0.97–2.06) 0.071 1.22 (0.59–2.52) 0.595 Low HDL cholesterol 1.33 (0.93–1.90) 0.118 1.64 (1.08–2.49) 0.020 0.84 (0.40–1.79) 0.659 ≤1 component vs. ≥2 1.39 (1.10–1.74) 0.005 1.28 (0.96–1.72) 0.090 1.48 (0.96–2.28) 0.075 components

Bold values merely corresponds with those with p values <0.05. Abbreviations: Adjusted Model: Adjusted for sex, age, smoking habit, educational level and toothbrushing frequency. CPI, Community Periodontal Index; NCEP-ATP, National Cholesterol Education Program Adult Treatment Panel III; IDF, International Diabetes Federation; HDL, high-density lipoprotein; OR, odds ratio; CI, confidence interval. with metabolically unhealthy subjects, particularly if they were the prevalence among subjects with CPI = 0 or CAL = 0–3 mm (for obese (OR = 1.78; 95% CI 1.28–2.49; p = 0.001, for CPI = 4; and CPI, 52.9% versus 24.1%, respectively; for CAL, 60.3% versus 29.1%, OR = 1.49; 95% CI 1.02- 2.20; p = 0.041 for CAL ≥ 6 mm), but no respectively). Similar observations were found for lipids, glycaemic association with metabolically healthy obese subjects was observed. levels and abdominal obesity. These findings are in agreement with previous investigations in other population-based studies, such as the NHANES III in the United States, where the prevalence of MetS 4 | DISCUSSION was two times higher among patients with moderate or severe periodontitis when compared with subjects with healthy or mild disease The results of the present study show a significant association be- (D'Aiuto et al., 2008). tween a worse periodontal condition and MetS, as defined using After adjusting for potential confounders (sex, age, smoking sta- widely accepted criteria. tus, educational level and toothbrushing frequency), high BP was In the present investigation, the prevalence of MetS was 8.5%, ac- the individual component of MetS that demonstrated the stron- cording to NCEP-ATP III definition and 16.2%, using the IDF criteria. gest association with severe periodontitis (defined either with CAL This higher prevalence using the definition of the IDF is in agreement or CPI), especially among women. A recent systematic review with with the findings from other epidemiological studies (Herath et al., meta-analysis showed an OR = 1.49 (95% CI 1.09–2.50) for the asso- 2018). In spite of these differences in MetS prevalence depending ciation between severe periodontitis and high BP (Munoz Aguilera on the two definitions used, the association between MetS and peri- et al., 2020). Moreover, recent evidence from a cross-sectional sur- odontitis was clear and consistent with both. Patients with severe vey among treated hypertensive adults have suggested that peri- periodontitis (defined as either CPI = 4 or CAL ≥ 6 mm) presented odontitis may be associated with unsuccessful antihypertensive a risk for suffering MetS more than two times higher than the one treatment, as periodontally healthy subjects presented a mean sys- for patients with a better periodontal condition. Furthermore, when tolic BP 2.3––3 mmHg lower, when compared with periodontitis pa- the components of the syndrome where evaluated separately, all tients (Pietropaoli et al., 2018). Further intervention trials are needed showed a clear tendency towards a significant association between to clarify the benefits of periodontal treatment upon BP, but on the a worse periodontal condition and high prevalence of cardiometa- basis of the available evidence, a positive effect may be expected. bolic abnormalities. As an example, the prevalence of high BP in the Besides high BP, the main contributors to the association be- group of patients diagnosed with CPI = 4 or CAL ≥ 6 mm doubled tween periodontitis and MetS reported in the literature are low 46 | MONTERO et al.

TABLE 5 Association between the different definitions of Metabolic Syndrome (MetS) and its individual components with severe periodontitis (defined as CAL ≥6 mm)

Total Male gender Female gender

p OR (95% CI) p value OR (95% CI) p value OR (95% CI) value

MetS (IDF definition) 1.37 (1.04–1.82) 0.027 1.29 (0.95–1.77) 0.105 1.77 (0.92–3.39) 0.085 Abdominal obesity 1.16 (0.89–1.51) 0.281 1.49 (1.11–2.00) 0.008 1.44 (0.83–2.51) 0.196 Hypertriglyceridaemia 1.17 (0.93–1.48) 0.188 1.02 (0.76–1.37) 0.898 1.75 (0.97–3.18) 0.065 High blood pressure 1.94 (1.49–2.53) <.001 1.90 (1.41–2.57) <.001 2.24 (1.29–3.88) 0.004 Hyperglycaemia 1.29 (0.96–1.74) 0.094 1.13 (0.81–1.58) 0.477 1.99 (1.04–3.81) 0.037 Low HDL cholesterol 1.33 (0.97–1.80) 0.073 1.57 (1.12–2.20) 0.009 0.61 (0.25–1.45) 0.260 ≤1 component vs ≥2 1.47 (1.13–1.90) 0.004 1.88 (1.31–2.68) 0.001 1.94 (1.13–3.31) 0.016 components MetS (NCEP-ATP III 1.57 (1.11–2.21) 0.010 1.29 (0.87–1.90) 0.209 3.63 (1.80–7.20) <.001 definition) Abdominal obesity 1.19 (0.93–1.52) 0.174 1.12 (0.81–1.55) 0.493 1.64 (0.94–2.85) 0.080 Hypertriglyceridaemia 1.17 (0.88–1.54) 0.276 0.99 (0.73–1.35) 0.960 2.18 (1.18–4.03) 0.013 High blood pressure 1.96 (1.51–2.54) <.001 1.92 (1.43–2.57) <.001 2.34 (1.35–4.06) 0.002 Hyperglycaemia 1.49 (1.03–2.15) 0.034 1.35 (0.89–2.04) 0.160 1.91 (0.87–4.17) 0.105 Low HDL cholesterol 1.35 (0.90–2.02) 0.148 1.63 (1.03–2.58) 0.038 0.88 (0.34–2.26) 0.789 ≤1 component vs. ≥2 1.57 (1.21–2.03) 0.001 1.75 (1.23–2.47) 0.002 2.09 (1.19–3.68) 0.011 components

TABLE 6 Association between periodontal measures (CPI and CAL) and body size phenotypes

CPI Code 4 CAL ≥ 6 mm

Crude model Adjusted model Crude model Adjusted model

Body size p phenotype OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) Value

MHNOa MHO 1.42 (1.00–2.03) 0.053 1.03 (0.71–1.51) 0.862 1.52 (1.01–2.29) 0.044 1.03 (0.67–1.59) 0.908 MNHNO 2.05 (1.60–2.63) <.001 1.31 (1.00–1.71) 0.047 2.52 (1.92–3.32) <.001 1.46 (1.09–1.96) 0.011 MNHO 3.05 (2.24–4.15) <.001 1.78 (1.28–2.49) 0.001 2.76 (1.92–3.97) <.001 1.49 (1.02–2.20) 0.041

Adjusted Model: Adjusted for sex, age and smoking habit. Bold values merely corresponds with those with p values <0.05. Abbreviations: CAL, clinical attachment level; CI, confidence interval; MHNO: Metabolically Healthy Non-Obese; MHO: Metabolically Healthy Obese; MNHNO: Metabolically Non-Healthy Non-Obese; MNHO: Metabolically Non-Healthy Obese. CPI, Community Periodontal Index; OR, odds ratio. aReference category. levels of HDL cholesterol and elevated fasting glucose (Alhabashneh populations, and using different measures of periodontitis (e.g. mean et al., 2015; D'Aiuto et al., 2008; Shimazaki et al., 2007; Tu et al., probing pocket depth, percentage of sites with CAL ≥ 4 mm), re- 2013). According to the IDF criteria, elevated fasting plasma glucose porting no significant differences in the periodontal condition of includes both impaired fasting glucose (i.e. prediabetes) and values subjects with or without prediabetes {Noack, 2000 #66}{Kowall, characteristic of diabetes mellitus. Previous studies have reported 2015 #63}. Similar findings from the WORALTH study have been controversial data regarding the association between periodontitis previously published by our research group, proving a significant as- and prediabetes, with some studies demonstrating a significant association between periodontitis and diabetes mellitus, but failing to sociation {Saito, 2004 #343}{Zadik, 2010 #798}{Choi, 2011 #135} demonstrate an association between periodontitis and prediabetes {Arora, 2014 #44}, but with other studies performed in European {Montero, 2019 #799}. In the present manuscript, when pooling both MONTERO ET AL. | 47 conditions under the term “Hyperglycaemia”, no significant associa- studies, with ORs ranging between 1.52 and 3.60 (Andriankaja et al., tion could be found with CPI code 4 or CAL ≥ 6 mm. 2010; Furuta et al., 2013; Tu et al., 2013). Our findings support this Abdominal obesity, despite being a crucial component in the gender differences, with significant higher ORs for the association definition of MetS, does not seem to play a key role in the asso- between MetS and periodontitis for female patients, according to ciation with periodontitis. Most cross-sectional studies on MetS both definitions. Some authors have hypothesized that female pa- patients report either no association between obesity and periodon- tients with MetS have a higher systemic inflammatory status (rep- titis (Alhabashneh et al., 2015; Benguigui et al., 2010; D'Aiuto et al., resented by higher levels of C-reactive protein, CRP) than males 2008), or associations of small magnitude and no statistically signif- (Saltevo et al., 2008), and that the protective role of sex hormones icant (Timonen et al., 2010). These findings are in agreement with against body fat distribution and insulin activity may be diminished the results of the present study. Furthermore, when patients where by the effect of certain pro-inflammatory cytokines (Alpizar & Spicer, categorized by body size phenotype, obesity alone, with no concom- 1994), which are usually in higher levels in severe periodontitis pa- itant cardiometabolic abnormalities (MHO), did not show association tients (Higashi et al., 2009). Further research is needed to elucidate with periodontitis, while subjects presenting altered metabolic pro- these mechanisms explaining the significant association between files showed a greater tendency towards a worse periodontal condi- MetS and periodontitis in women. tion, independently of being obese or with normal weight (MNHNO When assessing the evaluated oral health behavioural factors, and MNHO, respectively). These observations suggest the possible a potential protective role of regular toothbrushing was observed, role of the low-grade systemic inflammatory status reported in se- with approximately 10% less probability of being diagnosed with vere periodontitis as a contributor in the pathophysiology of cer- MetS among subjects reporting to brush their teeth twice a day. tain metabolic alterations, such as the increase in plasma glucose Similar observations have been made by other authors, reporting levels and abnormalities in the lipid profiles, even in the absence of an OR = 21.82 for suffering MetS when subjects reported tooth- large fat deposits. In any case, it should be noted that all anthro- brushing frequencies lower than twice per day (Thanakun et al., pometric measures (WC, BMI, etc.) are not necessarily indicative of 2014). The potential preventive role of oral hygiene on the onset of the amount of visceral fat, and therefore, in order to evaluate the metabolic disorders, together with the already established benefits potential association between adiposity and periodontitis, imaging of treating periodontitis in patients with diabetes mellitus (D'Aiuto techniques such as magnetic resonance should be used in the future. et al., 2018), advocate for, at least, a recommendation for regular The association between periodontitis and MetS seems to periodontal examinations and oral hygiene promotion in subjects be stronger in women, as reported in different population-based with MetS.

Subjects selected for the oral health examinaon (n = 5201) Subjects refusing to aend the oral health examinaon (n = 47)

Subjects included in the oral health examinaon (n = 5154)

Edentulous Subjects (n = 24)

Subjects receiving a periodontal examinaon (CPI and CAL analyses) (n = 5130) Subjects with missing data for MetS diagnosis (n = 777) Without waist circumference (n = 735) Without blood pressure (n = 609) Without triglycerides (n = 878) Subjects included in the ﬁnal analyses Without HDL (n = 869) Without FPG (n = 908) (n = 4353)

FIGURE 1 Flowchart of the subjects included/excluded in the WORALTH study and in the presence of Metabolic Syndrome (MetS) analysis. CAL, Clinical Attachment Level; CPI, Community Periodontal Index; FPG, Fasting Plasma Glucose); HDL, High-Density Lipoprotein 48 | MONTERO et al.

Benguigui, C., Bongard, V., Ruidavets, J.-B., Chamontin, B., Sixou, M., Different limitations in this investigation should be acknowl- Ferrières, J., & Amar, J. (2010). Metabolic syndrome, insulin resis- edged. Firstly, due to the cross-sectional nature of the present tance, and periodontitis: A cross-sectional study in a middle-aged study, the direction of the association, or a possible causative role, French population. Journal of Clinical Periodontology, 37(7), 601– cannot be established. Secondly, since the Spanish employed popu- 608. https://doi.org/10.1111/j.1600-051X.2010.01571.x Bizzarro, S., & Loos, B. G. (2019). The link between periodontitis and lation was the target group, only subjects aged between 16 and 65 erectile dysfunction: a review. British Dental Journal, 227(7), 599– were included, which may limit the external validity. Furthermore, 603. https://doi.org/10.1038/s41415-019-0724-6 some of the patients (n = 777) had to be excluded as they did not Cameron, A. J., Shaw, J. E., & Zimmet, P. Z. (2004). The metabolic syn- present enough data to perform a diagnosis of MetS. And thirdly, drome: prevalence in worldwide populations. Endocrinology and Metabolism Clinics of North America, 33(2), 351–375. https://doi. the periodontal assessment based on the WHO recommendation, org/10.1016/j.ecl.2004.03.005 consisting on a partial-mouth examination, may underestimate the Carasol, M., Llodra, J. C., Fernández-Meseguer, A., Bravo, M., García- levels of disease. Margallo, M. T., Calvo-Bonacho, E., Sanz, M., & Herrera, D. (2016). Despite the referred limitations of the present investigation, a Periodontal conditions among employed adults in Spain. Journal of Clinical Periodontology, 43(7), 548–556. https://doi.org/10.1111/ statistically significant association between metabolic syndrome jcpe.12558 and a worse periodontal condition has been observed in the work- D'Aiuto, F., Gkranias, N., Bhowruth, D., Khan, T., Orlandi, M., Suvan, J., & ing population in Spain, being high blood pressure the component Deanfield, J. E. (2018). Systemic effects of periodontitis treatment with the strongest role. This association was independent of BMI in patients with type 2 diabetes: a 12 month, single-centre, inves- tigator-masked, randomised trial. Lancet Diabetes Endocrinol., 6(12), and other confounders such as age, sex or toothbrushing frequency. 954–965. https://doi.org/10.1016/S2213-8587(1018)30038-X However, further research is needed through observational longitu- D'Aiuto, F., Sabbah, W., Netuveli, G., Donos, N., Hingorani, A. D., dinal studies or intervention trials in order to determine the possible Deanfield, J., & Tsakos, G. (2008). Association of the metabolic syn- causative relationship between MetS and periodontitis, as well as drome with severe periodontitis in a large U.S. population-based survey. , (10), the benefits derived from periodontal treatment in the cardiometa- Journal of Clinical Endocrinology and Metabolism 93 3989–3994. https://doi.org/10.1210/jc.2007-2522 bolic risk of MetS patients. de Pablo, P., Chapple, I. L., Buckley, C. D., & Dietrich, T. (2009). Periodontitis in systemic rheumatic diseases. Nature Reviews Rheumatology, 5(4), CONFLICT OF INTEREST 218–224. https://doi.org/10.1038/nrrheum.2009.28 The authors declare that there are no conflicts of interest related to Eckel, R. H., Grundy, S. M., & Zimmet, P. Z. (2005). The metabolic syndrome. Lancet, 365(9468), 1415–1428. https://doi.org/10.1016/ this research. S 0 1 4 0 -6736(05)66378 - 7 Friedewald, W. T., Levy, R. I., & Fredrickson, D. S. (1972). Estimation of DATA AVAILABILITY STATEMENT the concentration of low-density lipoprotein cholesterol in plasma, Data availability will be considered by the authors upon request. without use of the preparative ultracentrifuge. Clinical Chemistry, 18(6), 499–502. Furuta, M., Shimazaki, Y., Takeshita, T., Shibata, Y., Akifusa, S., Eshima, ORCID N., Kiyohara, Y., Ninomiya, T., Hirakawa, Y., Mukai, N., Nagata, M., Eduardo Montero https://orcid.org/0000-0003-2525-8529 & Yamashita, Y. (2013). Gender differences in the association be- Mariano Sanz https://orcid.org/0000-0002-6293-5755 tween metabolic syndrome and periodontal disease: the Hisayama Study. Journal of Clinical Periodontology, 40(8), 743–752. https://doi. David Herrera https://orcid.org/0000-0002-5554-2777 org/10.1111/jcpe.12119 Gorgel, S. N., Gorgel, A., & Sefik, E. (2014). Sexual function in male REFERENCES patients with metabolic syndrome and effective parame- Alhabashneh, R., Khader, Y., Herra, Z., Asa'ad, F., & Assad, F. (2015). The ters on erectile dysfunction. International Brazilian Journal of association between periodontal disease and metabolic syndrome Urology, 40(1), 56–61. https://doi.org/10.1590/S1677-5538. among outpatients with diabetes in Jordan. Journal of Diabetes IBJU.2014.01.08 & Metabolic Disorders, 14(67), eCollection 42015. https://doi. Han, T. S., & Lean, M. E. (2016). A clinical perspective of obesity, met- org/10.1186/s40200-40015-40192-40208 abolic syndrome and cardiovascular disease. JRSM Cardiovascular Alpizar, E., & Spicer, L. J. (1994). Effects of interleukin-6 on prolifera- Disease, 5, 2048004016633371. https://doi.org/10.1177/20480 tion and follicle-stimulating hormone-induced estradiol production 04016633371 by bovine granulosa cells in vitro: Dependence on size of follicle. Herath, H. M. M., Weerasinghe, N. P., Weerarathna, T. P., & Amarathunga, Biology of Reproduction, 50(1), 38–43. https://doi.org/10.1095/biolr A. (2018). A comparison of the prevalence of the metabolic syn- eprod1050.1091.1038 drome among Sri Lankan patients with type 2 diabetes mellitus Andriankaja, O. M., Sreenivasa, S., Dunford, R., & DeNardin, E. (2010). using WHO, NCEP-ATP III, and IDF definitions. International Journal Association between metabolic syndrome and periodontal dis- of Chronic Diseases, 2018:7813537. eCollection 7812018. https:// ease. Australian Dental Journal, 55(3), 252–259. https://doi. doi.org/10.1155/2018/7813537 org/10.1111/j.1834-7819.2010.01231.x Herrera, D., Meyle, J., Renvert, S., & Jin, L. (2018). White paper on preven- Beaglehole, R., Bonita, R., Horton, R., Adams, C., Alleyne, G., Asaria, tion and management of periodontal diseases for oral health and gen- P., Baugh, V., Bekedam, H., Billo, N., Casswell, S., Cecchini, M., eral health. FDI Global Periodontal Health Project Task Team Genève, Colagiuri, R., Colagiuri, S., Collins, T., Ebrahim, S., Engelgau, M., Switzerland. FDI World Dental Federation. Galea, G., Gaziano, T., Geneau, R., … Watt, J. (2011). Priority ac- Higashi, Y., Goto, C., Hidaka, T., Soga, J., Nakamura, S., Fujii, Y., Hata, T., tions for the non-communicable disease crisis. Lancet, 377(9775), Idei, N., Fujimura, N., Chayama, K., Kihara, Y., & Taguchi, A. (2009). 1438–1447. https://doi.org/10.1016/S0140-6736(11)60393-0 Oral infection-inflammatory pathway, periodontitis, is a risk factor MONTERO ET AL. | 49

for endothelial dysfunction in patients with coronary artery dis- control among us hypertensive adults. Hypertension, 72(6), 1365– ease. Atherosclerosis, 206(2), 604–610. https://doi.org/10.1016/j. 1373. 1310.1161/HYPERTENSIONAHA.1118.11528 atherosclerosis.2009.03.037 Pihlstrom, B. L., Michalowicz, B. S., & Johnson, N. W. (2005). Periodontal Instituto Nacional de Estadística. (2008). Encuesta de Población Activa 2º diseases. Lancet, 366(9499), 1809–1820. https://doi.org/10.1016/ Trimestre (EPA 2008TII). S 0 1 4 0 -6736(05)67728 - 8 International Diabetes Federation (2015). The IDF consensus worldwide Prasad, D. S., Kabir, Z., Dash, A. K., & Das, B. C. (2012). Prevalence and risk definition of the metabolic syndrome. Retrieved from https://www. factors for metabolic syndrome in Asian Indians: A community study idf.org/e-libra r y / c o n s e n s u s - s t a t e m e n t s / 6 0 - i d f c o n s e n s u s - w o r l d from urban Eastern India. Journal of Cardiovascular Disease Research, wide-definitiono f-the-metabolic-syndrome.html 3(3), 204–211. https://doi.org/10.4103/0975-3583.98895 Karelis, A. D., St-Pierre, D. H., Conus, F., Rabasa-Lhoret, R., & Poehlman, Reyes, L., Herrera, D., Kozarov, E., Roldan, S., & Progulske-Fox, A. (2013). E. T. (2004). Metabolic and body composition factors in subgroups Periodontal bacterial invasion and infection: contribution to ath- of obesity: what do we know? Journal of Clinical Endocrinology erosclerotic pathology. Journal of Clinical Periodontology, 40(Suppl and Metabolism, 89(6), 2569–2575. https://doi.org/10.1210/ 14), S30–50. https://doi.org/10.1111/jcpe.12079 jc.2004-0165 Sacks, D. B., Arnold, M., Bakris, G. L., Bruns, D. E., Horvath, A. R., Kassebaum, N. J., Bernabe, E., Dahiya, M., Bhandari, B., Murray, C. J., & Kirkman, M. S., Lernmark, A., Metzger, B. E., & Nathan, D. M. (2011). Marcenes, W. (2014). Global burden of severe periodontitis in 1990– Guidelines and recommendations for laboratory analysis in the di- 2010: a systematic review and meta-regression. Journal of Dental agnosis and management of diabetes mellitus. Clinical Chemistry, Research, 93(11), 1045–1053. https://doi.org/10.1177/00220 57(6), e1–e47. https://doi.org/10.1373/clinchem.2010.161596 34514552491 Saltevo, J., Vanhala, M., Kautiainen, H., Kumpusalo, E., & Laakso, M. LaMonte, M. J., Williams, A. L. M., Genco, R. J., Andrews, C. A., Hovey, K. (2008). Gender differences in C-reactive protein, interleukin-1 M., Millen, A. E., Browne, R. W., Trevisan, M., & Wactawski-Wende, receptor antagonist and adiponectin levels in the metabolic syn- J. (2014). Association between metabolic syndrome and peri- drome: a population-based study. Diabetic Medicine, 25(6), 747– odontal disease measures in postmenopausal women: the Buffalo 750. https://doi.org/10.1111/j.1464-5491.2008.02440.x OsteoPerio study. Journal of Periodontology, 85(11), 1489–1501. Sánchez Chaparro, M. A., Calvo Bonacho, E., González Quintela, A., https://doi.org/10.1902/jop.2014.140185 Cabrera, M., Sáinz, J. C., Fernández-Labander, C., Quevedo- Martinez-Larrad, M. T., Corbaton-Anchuelo, A., Fernandez-Perez, C., Aguado, L., Gelpi, J. A., Fernández Meseguer, A., Brotons, C., & Lazcano-Redondo, Y., Escobar-Jimenez, F., & Serrano-Rios, M. Icaria Study Group (2011). High cardiovascular risk in Spanish work- (2016). Metabolic syndrome, glucose tolerance categories and ers. Nutrition, Metabolism & Cardiovascular Diseases, 21(4), 231–236. the cardiovascular risk in Spanish population. Diabetes Research https://doi.org/10.1016/j.numecd.2009.10.001 and Clinical Practice, 114, 23–31. https://doi.org/10.1016/j.diabr Sanchez-Chaparro, M.-A., Calvo-Bonacho, E., Gonzalez-Quintela, A., es.2016.02.003 Fernandez-Labandera, C., Cabrera, M., Sainz, J.-C., Fernandez- Misra, A., & Khurana, L. (2008). Obesity and the metabolic syndrome Meseguer, A., Banegas, J. R., Ruilope, L.-M., Valdivielso, P., & in developing countries. Journal of Clinical Endocrinology and Roman-Garcia, J., & Ibermutuamur Cardiovascular Risk Assessment Metabolism, 93(11 Suppl 1), S9–S30. https://doi.org/10.1210/ Study, G (2008). Occupation-related differences in the prevalence jc.2008-1595 of metabolic syndrome. Diabetes Care, 31(9), 1884–1885. https:// Morita, T., Yamazaki, Y., Mita, A., Takada, K., Seto, M., Nishinoue, N., doi.org/10.2337/dc08-0431 Sasaki, Y., Motohashi, M., & Maeno, M. (2010). A cohort study on Sánchez-Chaparro, M. A., Román-García, J., Calvo-Bonacho, E., Gómez- the association between periodontal disease and the development Larios, T., Fernández-Meseguer, A., Sáinz-Gutiérrez, J. C., Cabrera- of metabolic syndrome. Journal of Periodontology, 81(4), 512–519. Sierra, M., García-García, A., Rueda-Vicente, J., Gálvez-Moraleda, https://doi.org/10.1902/jop.2010.090594 A., & González-Quintela, A. (2006). Prevalence of cardiovascular Munoz Aguilera, E., Suvan, J., Buti, J., Czesnikiewicz-Guzik, M., Barbosa risk factors in the Spanish working population. Revista Espanola De Ribeiro, A., Orlandi, M., & D'Aiuto, F. (2020). Periodontitis is asso- Cardiologia, 59(5), 421–430. ciated with hypertension: a systematic review and meta-analysis. Schenkein, H. A., & Loos, B. G. (2013). Inflammatory mechanisms link- Cardiovascular Research, 116(1), 28–39. https://doi.org/10.1093/ ing periodontal diseases to cardiovascular diseases. Journal of cvr/cvz1201 Periodontology, 84(4 Suppl), S51–69. https://doi.org/10.1902/ National Cholesterol Education Program Expert Panel on Detection, E., jop.2013.134006 & Treatment of High Blood Cholesterol in, A (2002). Third Report of Shimazaki, Y., Saito, T., Yonemoto, K., Kiyohara, Y., Iida, M., & Yamashita, the National Cholesterol Education Program (NCEP) Expert Panel Y. (2007). Relationship of metabolic syndrome to periodontal dis- on Detection, Evaluation, and Treatment of High Blood Cholesterol ease in Japanese women: the Hisayama Study. Journal of Dental in Adults (Adult Treatment Panel III) final report. Circulation, Research, 86(3), 271–275. https://doi.org/10.1177/1544059107 106(25), 3143–3421. 08600314 Nibali, L., Tatarakis, N., Needleman, I., Tu, Y. K., D'Aiuto, F., Rizzo, M., & Sora, N. D., Marlow, N. M., Bandyopadhyay, D., Leite, R. S., Slate, E. H., Donos, N. (2013). Clinical review: Association between metabolic & Fernandes, J. K. (2013). Metabolic syndrome and periodontitis syndrome and periodontitis: a systematic review and meta-analy- in Gullah African Americans with type 2 diabetes mellitus. Journal sis. Journal of Clinical Endocrinology and Metabolism, 98(3), 913–920. of Clinical Periodontology, 40(6), 599–606. https://doi.org/10.1111/ https://doi.org/10.1210/jc.2012-3552 jcpe.12104 Paraskevas, S., Huizinga, J. D., & Loos, B. G. (2008). A systematic review Taylor, J. J., Preshaw, P. M., & Lalla, E. (2013). A review of the evidence for and meta-analyses on C-reactive protein in relation to periodon- pathogenic mechanisms that may link periodontitis and diabetes. titis. Journal of Clinical Periodontology, 35(4), 277–290. https://doi. Journal of Clinical Periodontology, 40(Suppl 14), S113–134. https:// org/10.1111/j.1600-051X.2007.01173.x doi.org/10.1111/jcpe.12059 Petersen, P. E., & Ogawa, H. (2012). The global burden of periodon- Thanakun, S., Watanabe, H., Thaweboon, S., & Izumi, Y. (2014). tal disease: towards integration with chronic disease preven- Association of untreated metabolic syndrome with moderate to tion and control. Periodontology 2000, 60(1), 15–39. https://doi. severe periodontitis in Thai population. Journal of Periodontology, org/10.1111/j.1600-0757.2011.00425.x 85(11), 1502–1514. 1510.1902/jop.2014.140105 Pietropaoli, D., Del Pinto, R., Ferri, C., Wright, J. T. Jr, Giannoni, M., Timonen, P., Niskanen, M., Suominen-Taipale, L., Jula, A., Knuuttila, M., Ortu, E., & Monaco, A. (2018). Poor oral health and blood pressure & Ylostalo, P. (2010). Metabolic syndrome, periodontal infection, 50 | MONTERO et al.

and dental caries. Journal of Dental Research, 89(10), 1068–1073. Zadik, Y., Bechor, R., Galor, S., Justo, D., & Heruti, R. J. (2009). 1010.1177/0022034510376542 Erectile dysfunction might be associated with chronic peri- Tonetti, M. S., Jepsen, S., Jin, L., & Otomo-Corgel, J. (2017). Impact odontal disease: two ends of the cardiovascular spectrum. of the global burden of periodontal diseases on health, nutri- Journal of Sexual Medicine, 6(4), 1111–1116. https://doi. tion and wellbeing of mankind: A call for global action. Journal of org/10.1111/j.1743-6109.2008.01141.x Clinical Periodontology, 44(5), 456–462. https://doi.org/10.1111/ Zuk, A., Quinonez, C., Lebenbaum, M., & Rosella, L. C. (2017). The jcpe.12732 association between undiagnosed glycaemic abnormalities Tu, Y. K., D'Aiuto, F., Lin, H. J., Chen, Y. W., & Chien, K. L. (2013). and cardiometabolic risk factors with periodontitis: results Relationship between metabolic syndrome and diagnoses of peri- from 2007–2009 Canadian Health Measures Survey. Journal of odontal diseases among participants in a large Taiwanese cohort. Clinical Periodontology, 44(2), 132–141. https://doi.org/10.1111/ Journal of Clinical Periodontology, 40(11), 994–1000. https://doi. jcpe.12684 org/10.1111/jcpe.12157 Valdivielso, P., Sanchez-Chaparro, M. A., Calvo-Bonacho, E., Cabrera- Sierra, M., Sainz-Gutierrez, J. C., Fernandez-Labandera, C., SUPPORTING INFORMATION Fernández-Meseguer, A., Quevedo-Aguado, L., Ruiz Moraga, M., Additional supporting information may be found online in the Gálvez-Moraleda, A., González-Quintela, A., & Roman-Garcia, J. Supporting Information section. & ICARIA (Ibermutuamur CArdiovascular RIsk Assesment) study group (2009). Association of moderate and severe hypertri- glyceridemia with obesity, diabetes mellitus and vascular disease in the Spanish working population: results of the ICARIA study. Atherosclerosis, 207(2), 573–578. https://doi.org/10.1016/j.ather How to cite this article: Montero E, Molina A, Carasol M, et al. osclerosis.2009.05.024 The association between metabolic syndrome and WHO (1997). Oral Health Surveys: Basic methods, 4th ed. World Health periodontitis in Spain: Results from the Organization. WORALTH (Workers’ Wildman, R. P., Muntner, P., Reynolds, K., McGinn, A. P., Rajpathak, S., ORAL healTH) Study. J Clin Periodontol. 2021;48:38–50. https:// Wylie-Rosett, J., & Sowers, M. R. (2008). The obese without car- doi.org/10.1111/jcpe.13391 diometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US population (NHANES 1999–2004). Archives of Internal Medicine, 168(15), 1617–1624. https://doi. org/10.1001/archinte.168.15.1617 Received: 1 April 2020 | Revised: 16 September 2020 | Accepted: 1 October 2020 DOI: 10.1111/jcpe.13386

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

A quantitative bias analysis to assess the impact of unmeasured confounding on associations between diabetes and periodontitis

Talal S. Alshihayb1,2 | Elizabeth A. Kaye1 | Yihong Zhao1,3 | Cataldo W. Leone1 | Brenda Heaton1

1Boston University Henry M. Goldman School of Dental Medicine, Boston, MA, Abstract USA Aim: To investigate unmeasured confounding in bidirectional associations between 2College of Dentistry, King Saud bin Abdulaziz University for Health Sciences, periodontitis and diabetes using quantitative bias analysis. Riyadh, Saudi Arabia Methods: Subsamples from the Veterans Affairs Dental Longitudinal Study were se- 3 Center of Alcohol and Substance lected. Adjusted for known confounders, we used Cox proportional hazards mod- Use Studies, Department of Applied Psychology, Graduate School of Applied els to estimate associations between pre-existing clinical periodontitis and incident and Professional Psychology, Rutgers Type II Diabetes (n = 672), and between pre-existing diabetes and incident severe University, Piscataway, NJ, USA periodontitis (n = 521), respectively. Hypothetical confounders were simulated into Correspondence the dataset using Bernoulli trials based on pre-specified distributions of confounders Talal S. Alshihayb, Boston University Henry M. Goldman School of Dental within categories of each exposure and outcome. We calculated corrected hazard rd Medicine, 560 Harrison Avenue, 3 floor, ratios (HR) over 10,000 bootstrapped samples. Rm 324I, Boston, MA 02118, USA. Email: [email protected] Results: In models using periodontitis as the exposure and incident diabetes as the Funding information outcome, adjusted HR = 1.21 (95% CI: 0.64–2.30). Further adjustment for simulated This is a secondary analysis of a de- identified dataset from the Dental confounders positively associated with periodontitis and diabetes greatly attenuated longitudinal study, which was supported the association or explained it away entirely (HR = 1). In models using diabetes as by the Veterans Affairs administration. the exposure and incident periodontitis as the outcome, adjusted HR = 1.35 (95% CI: 0.79–2.32). After further adjustment for simulated confounders, the lower bound of the simulation interval never reached the null value (HR ≥ 1.03). Conclusions: Presence of unmeasured confounding does not explain observed associations between pre-existing diabetes and incident periodontitis. However, presence of weak unmeasured confounding eliminated observed associations between pre-existing periodontitis and incident diabetes. These results clarify the bidirectional periodontitis-diabetes association.

KEYWORDS bias, confounding factors, diabetes mellitus, periodontal disease, periodontitis

wileyonlinelibrary.com/journal/jcpe | 51 J Clin Periodontol. 2021;48:51–60. 52 | ALSHIHAYB et al.

1 | INTRODUCTION Clinical Relevance Associations between periodontitis and various systemic diseases, Scientific rationale for study: Risk factor epidemiology including cardiovascular disease, diabetes and rheumatoid arthritis, supports the possible bidirectional association between have been identified (Linden et al., 2013). Despite extensive work periodontitis and diabetes, underscoring the role of inflam- in the area, the extent to which the observed associations may be mation in both disease processes. The hypothesized causal accounted for, in part or in whole, by potential unmeasured or un- pathways may include a causal factor common to both dis- recognized confounders remains unclear. The association between ease conditions, yet to be delineated, thereby introducing periodontitis and diabetes mellitus is particularly interesting in this confounding as a possible alternative explanation. regard, as it has been documented to be bidirectional in nature Principal findings: We report on a quantitative bias analysis (Chavarry et al., 2009; Demmer et al., 2012; Expert Committee demonstrating that the pre-existing periodontitis and inci- on the Diagnosis & Classification of Diabetes Mellitus, 2003; dent diabetes association may be vulnerable to confound- Khader et al., 2006; Löe, 1993; Mealey & Ocampo, 2007; Polak & ing, whereas the opposite association be less so. Shapira, 2018; Salvi et al., 2008; Sanz et al., 2018; Winning et al., Practical implications: Our results call into question the 2017). Periodontitis and diabetes are both characterized by inflam- presence of a truly causal bidirectional association be- matory processes, the effects of which could affect the develop- tween periodontitis and diabetes. ment and/or progression of the other condition (Preshaw et al., 2012). Furthermore, it has been shown that periodontal treatment among diabetics may improve their metabolic control (Madianos & Koromantzos, 2018; Simpson et al., 2010, 2015). is implemented by independently simulating the associations be- Observational studies of periodontitis and diabetes, as well as tween hypothesized unmeasured confounders and the empirically other systemic diseases, may be explained by systematic errors such measured exposure and the outcome variables and subsequently as confounding, including residual confounding and/or unmeasured controlling for the simulated confounders in the empirical expo- confounding. For example, it has been suggested that observed sure-disease association. We applied QBA methods to the empirical associations of periodontitis and cardiovascular disease may be bidirectional associations between periodontitis and diabetes. Our explained by various sources of confounding including behaviours aim was to identify the potential role of unmeasured confounding in such as smoking and shared genes in genome wide association stud- the observed bidirectional associations. ies (Aarabi et al., 2017; Hujoel et al., 2000; Joshipura et al., 1998). Such work has highlighted the critical importance of systematically identifying potential confounders and accounting for them in re- 2 | METHODS gression analyses of oral-systemic disease associations. One valu- able approach to aid in selecting and appropriately controlling for The Veterans Affairs (VA) Dental Longitudinal Study (DLS; Kapur confounding is the use of directed acyclic graphs (DAGs; Greenland et al., 1972) is a subset of the VA Normative Aging Study (NAS), an et al., 1999). A key goal of using DAGs is to identify candidate con- ongoing closed-panel cohort study of ageing and health. At the NAS founding variables based on the hypothesized causal associations baseline, in the early 1960s, 2,280 adult men living in the Boston, with the studied exposure and outcome and to statistically control Massachusetts (USA), metropolitan area were enrolled (Bell et al., for them (Akinkugbe et al., 2016; Greenland et al., 1999; Merchant 1966). Men had to be free of chronic medical conditions for entry & Pitiphat, 2002). into the NAS. The cohort was almost entirely non-Hispanic white In the case of observed bidirectional associations between peri- (98%), reflective of the Boston area population at the time. In 1968, odontitis and diabetes, it is possible that there exists a common, 1,231 NAS participants self-selected into the DLS (Kapur et al., though yet unrecognized, underlying factor that causes these ob- 1972). DLS participants have returned at approximately three-year served associations. Plausibly, there may be a common inflammatory intervals for comprehensive medical and dental exams. While the pathway that results in the development and/or progression of both majority of enrolees were veterans, it is important to note that conditions. Thus, both the existence and the strength of the ob- they are not patients of the VA health care system. Throughout the served epidemiologic associations may be explainable by such con- DLS they have received their medical and dental care in the private founding. This should not be surprising, as our understanding of the sector. causal pathways between periodontitis and diabetes remains limited. Importantly, identifying the presence and role of an unknown confounder may provide alternative explanations for the observed 2.1 | Periodontitis ascertainment associations between periodontitis and diabetes. Novel epidemiologic techniques, such as quantitative bias anal- Clinical periodontal examinations, including measures of probing ysis (QBA), have been developed to assess the potential impact of pocket depth at multiple sites per tooth, were conducted beginning unmeasured confounding (Lash et al., 2009, 2014). Briefly, QBA at the DLS baseline (1968–1971). However, measurement of clinical ALSHIHAYB et al. | 53 attachment loss was only begun in 1981, at the start of the fifth tri- assumes that smoking effects on chronic disease take the form of an ennial DLS examination cycle. Detailed periodontal measurements exponentially decreasing function. CSI was calculated using the half- were collected from 1981-2009 by two examiners who were reli- life parameter for chronic periodontitis, which was estimated to be able and consistent (Feldman et al., 1982). For our analyses, we used 1.5 years (Dietrich & Hoffmann, 2004). Body Mass Index (BMI) was these DLS periodontal measurements and applied the case defini- calculated from weight and height measurements, obtained by the tions for periodontitis jointly developed in 2007 by the Centers for NAS using calibrated instruments, as BMI = weight (kg)/height(m)2. Disease Control and Prevention (CDC) and the American Academy Age, CSI and BMI were analysed as continuous variables. of Periodontology (AAP) (Page & Eke, 2007). In the analyses where periodontitis was evaluated as an exposure (i.e. pre-existing periodontitis as the exposure and its association with incident diabetes 2.4 | Statistical analyses as the outcome), periodontal status at the person-level was dichotomized into either present (moderate or severe periodontitis) or ab- All statistical analyses were carried out using SAS 9.4. Crude (unad- sent (mild or no periodontitis). In contrast, in the analyses where justed) and adjusted Cox proportional hazards models were fit with periodontitis was evaluated as an outcome, only persons who de- time-varying exposures and baseline covariates using follow-up time veloped “severe” periodontitis were considered to be incident cases. as a time scale. Adjusted Cox models with time-varying covariates, a form of non-proportional hazards, were also evaluated (Allison, 2010). Models with death from diabetes as a competing event were 2.2 | Diabetes ascertainment also estimated. Hazard ratios with their corresponding 95% confidence intervals were calculated. Loss to follow-up was minimal in The existing NAS medical records served as the primary source for this cohort and censoring over the follow-up period was primarily the diagnosis of diabetes. Diabetes status was ascertained during due to participants’ death (Heaton et al., 2014). Missing data on peri- each periodic NAS medical examination interviews based on self- odontitis, diabetes and covariates were uncommon (≤40 individuals). report. In analyses where incident diabetes was the outcome of in- Imputation of missing data was conducted by assuming mid-point terest, participants were also considered to have developed diabetes change, taking the average of scores immediately preceding and fol- if it was listed on their death certificate as a primary or secondary lowing the missing exam, or taking the same value as the last exam. cause of death.

2.5 | Analysis of the association between pre- 2.3 | Covariate ascertainment and control existing periodontitis (exposure) and incident type II diabetes (outcome) We developed two directed acyclic graphs (DAGs), using DAGitty (Textor et al., 2011), to guide our analyses. Figure 1 depicts the hy- DLS participants (N = 672) who were free of diabetes at the analysis pothesized causal associations between the exposure and outcome. baseline were identified. Participants also had to have at least eight The DAG labelled “A,” depicts the hypothesized causal association teeth present (excluding third molars) at their first periodontal exam between pre-existing periodontitis (exposure) and incident diabe- and to have data on diabetes status over the study follow-up period tes (outcome), while the DAG labelled “B,” depicts the hypothesized (1981–2009). Participants were excluded if they were completely causal associations between pre-existing diabetes (exposure) and in- missing data on any covariate across all examination time points. cident periodontitis (outcome). Based on the causal graph structure, Participants whose periodontal status could be categorized as mod- we identified a parsimonious set of covariates eligible for control erate or severe periodontitis were considered to have pre-existing in each of the analyses. These covariates were age, gender, socio- periodontitis (the exposure) for the purpose of the analysis. economic status, smoking, obesity and race/ethnicity. However, as The event of interest was the first diabetic diagnosis. Censoring the DLS restricted study enrolment to males, and due to the cohort was defined as any person who was lost to follow-up, died from rea- being 98% non-Hispanic white, there was no need to control for sons other than diabetes or completed the follow-up period without gender or race/ethnicity. developing diabetes (administrative censoring). If a DLS participant In our analyses, we used baseline values of participants’ highest had more than a seven-year gap between any study follow-up cy- attained education level as a proxy to control for socioeconomic sta- cles, he was artificially censored at seven years. Individuals were tus, and dichotomized it as college graduate or higher versus less than followed-up from their first periodontal examination until they de- college graduate. Data on tobacco smoking history, including its du- veloped diabetes or were censored. The date of the NAS examina- ration, intensity and time since quitting smoking, were collected via tion at which a new diabetes diagnosis was recorded was used as questionnaire at baseline and at subsequent follow-up examinations. the time point for a case of incident diabetes, except for those cases These data were used to calculate the comprehensive smoking index where diabetes was first identified via death certificate and we used (CSI) (Dietrich & Hoffmann, 2004; Leffondre et al., 2006). The CSI date of death as the date of incident diabetes. 54 | ALSHIHAYB et al.

FIGURE 1 (a) A causal diagram showing the hypothesized pre-existing periodontitis exposure and incident diabetes outcome association along with variables affecting periodontitis, diabetes or both. Common causes including age, gender, socioeconomic status, race/ethnicity, smoking and obesity affect both periodontitis and diabetes and thus need to be controlled to eliminate confounding. Variables such as inflammation, tooth loss and prediabetes represent mediators on the causal pathway from periodontitis to diabetes and thus, do not need be controlled. Family history affects only diabetes and thus, does not need to be controlled. (b) A causal diagram showing the hypothesized pre- existing diabetes exposure and incident periodontitis outcome association, in which, inflammation is the only mediator in this case. If there was an unmeasured variable that is not yet known to be a common cause of both periodontitis and diabetes and is not in the causal pathway, it could cause residual confounding even after appropriately adjusting for age, gender, socioeconomic status, race/ethnicity, smoking and obesity

2.6 | Analysis of the association between pre- Human Subjects Research approval was obtained from the existing diabetes (exposure) and incident periodontitis Institutional Review Boards at Boston University Medical Campus (outcome) and the Veterans Affairs Boston Healthcare System.

DLS participants (N = 521) who were free of severe periodontitis, had at least 8 teeth present (excluding third molars) at their 3 | RESULTS first periodontal exam, had at least 2 periodontal examinations over the study follow-up (1981–2009) and had data on all co- 3.1 | Association of pre-existing periodontitis variates for at least one DLS examination cycle were identified. (exposure) and incident type II diabetes (outcome) Individuals were followed-up from their first periodontal examination until they developed severe periodontitis or were cen- Table 1 (part A) shows the baseline distribution of potential con- sored. Censoring occurred at date of death, date of last exam if founders by periodontitis status. Participants presenting with mod- they completed the follow-up period without developing severe erate or severe periodontitis were older, less educated, had fewer periodontitis, or were lost to follow-up. Participants were also teeth and more likely to smoke and be obese than participants with censored if they experienced tooth loss such that fewer than two mild or no periodontitis. The median follow-up time was 14.0 years teeth were under observation (i.e. the person was no longer con- (range: 1.0–33.6). The difference in per cent of participants lost to sidered to be “at-risk” for severe periodontitis by the study's case follow-up between periodontitis groups was small (3.5%). definition). Table 2 (part A) shows the results of the time-dependent Cox proportional hazards models. In the unadjusted model, those with moderate or severe periodontitis had 1.46 times the hazard (95% 2.7 | Quantitative bias analysis CI: 0.79, 2.71) of developing type II diabetes compared to those with mild or no periodontitis. Adjustment for either baseline covariates or We conducted a quantitative bias analysis (QBA) to assess the time-dependent covariates, resulted in attenuation of the estimates potential impact of simulated unmeasured confounders on the to 1.21 (95% CI: 0.64, 2.30) or 1.33 (95% CI: 0.71, 2.52), respectively. association between pre-existing periodontitis and incident dia- Using a competing events model did not change the hazard ratio by betes, as well as the potential impact of simulated unmeasured more than 10% (data not shown). confounders on the association between pre-existing diabetes Table 3 (part A) shows the results of the QBA for the association and incident periodontitis. We applied standard QBA methods, between pre-existing periodontitis and incident type II diabetes. outlined in detail by Lash et al. (2009), Lash et al. (2014). Details We examined the impact of adjusting for baseline covariates after on the step-by-step process are available in the Appendix S1 to simulating a single confounder with different specifications on the this report. hazard of developing type II diabetes for each exposure group. We ALSHIHAYB et al. | 55

TABLE 1 Baseline distributions of covariates for the pre-existing periodontitis exposure and incident diabetes outcome association cohort (1A) and the pre-existing diabetes exposure and incident periodontitis outcome association cohort (1B)

1A: Pre-existing periodontitis exposure and incident diabetes outcome association

Severe/moderate periodontitis Mild/none

Variables N = 349 (51.9%) N = 323 (48.1%)

Mean age (SD) 61.3 (8.3) 57.7 (7.4) Socioeconomic status (education) % ≥College graduate 104 (29.8%) 112 (34.7%)

1B: Pre-existing diabetes exposure and incident periodontitis outcome association

Diabetic Non-diabetic

Variables N = 14 (2.7%) N = 507 (97.3%)

Mean age (SD) 62.3 (5.0) 58.8 (7.8) Socioeconomic status (education) % ≥College graduate 5 (35.7%) 176 (34.7%)

Abbreviations: BMI, body mass index; CSI, comprehensive smoking index; CVD, cardiovascular disease;SD, standard deviation. aHypertension was defined as having systolic blood pressure ≥130 mmHg, diastolic blood pressure ≥80 mmHg, or reporting taking a medication primarily for high blood pressure. bCVD was defined as a self-reported having coronary heart disease or stroke. found that a simulated confounder that had moderate-to-strong associated with periodontitis but also negatively associated with dia- positive associations with diabetes (e.g. with OR = 1.3–2), and that betes, or vice-versa, then the hazard ratio increased to as high as 1.8. was also positively associated with periodontitis, could substantially weaken the association between moderate or severe periodontitis and incident type II diabetes, or totally explain away the associa- 3.2 | Association of pre-existing diabetes tion (i.e. adjusted HR = 1). Similarly, a simulated confounder that (exposure) and incident periodontitis (outcome) had moderate-to-strong negative associations with diabetes (e.g. with OR = 0.5–0.8) and that was also negatively associated with Table 1 (part B) shows the baseline distribution of confounders by periodontitis, could similarly weaken or totally explain the non-null diabetes groups. Diabetics were older and more likely to be smokers association. In contrast, if the simulated confounder was positively and be obese than non-diabetics, but had the same education. Both 56 | ALSHIHAYB et al. groups had the same number of teeth at baseline, on average. The 4 | DISCUSSION median follow-up time was 8.5 years (range: 1.2–27.4). The difference in the per cent of participants lost to follow-up between diabe- The association between periodontitis and diabetes is complex and tes groups was small (diabetics had more loss by 5.0%). the role of confounders on potential causal relationships is not well Table 2 (part B) shows the time-dependent models for devel- understood. Similar concerns have also been raised about the asso- oping severe periodontitis. In the unadjusted model, diabetics had ciations of periodontitis with other systemic diseases, such as cardi- 1.45 times greater hazard of developing severe periodontitis com- ovascular disease (Aarabi et al., 2017; Hujoel et al., 2000; Joshipura pared to non-diabetics (95% CI: 0.86–2.45). After adjusting for either et al., 1998). The particular bidirectional nature of periodontitis-di- baseline covariates or time-dependent covariates, the hazard ratios abetes associations adds additional challenges to achieving a fuller decreased to 1.35 (95% CI: 0.79–2.32) or 1.32 (95% CI: 0.77–2.26), understanding of the role of confounding. In our study, we had respectively. the unique opportunity to explore confounding by using an exist- Table 3 (part B) shows the results of the QBA for the association ing longitudinal dataset that had detailed information on both the between pre-existing diabetes and incident severe periodontitis. periodontal and systemic health status of over 1,000 participants, A simulated confounder that had strong positive associations with many with multi-decade follow-up times. In addition, we used QBA, severe periodontitis (e.g. OR = 1.8–2), was at least 80% prevalent a method for systematically simulating the associations between among diabetics, and at most 20% prevalent among non-diabetics, hypothesized unmeasured confounders and the exposure and the weakened the association between diabetes and severe periodonti- outcome. The work reported here, the first to apply QBA to the tis (i.e. decreased HR) but never totally explained it. Similarly, a sim- study of an important oral-systemic disease association, has yielded ulated confounder that had strong negative associations with severe important insights. periodontitis (e.g. OR = 0.5–0.6), was at most 20% prevalent among In our analyses, when pre-existing diabetes was the exposure, diabetics, and at least 80% prevalent among non-diabetics, could its association with periodontitis was more robust to the effects of also weaken but never totally explain the association. In contrast, unmeasured confounding under all simulation scenarios, as com- if the simulated confounder was positively associated with severe pared to when pre-existing periodontitis was the exposure and periodontitis but negatively associated with diabetes, or vice-versa, incident type II diabetes was the outcome. Importantly, no matter then the hazard ratio was increased and could be as high as 4.4. how strong the unmeasured confounding was, it could never totally

TABLE 2 Crude and adjusted Cox proportional hazards models for the pre-existing periodontitis exposure and incident diabetes outcome association (2A) and the pre-existing diabetes exposure and incident periodontitis outcome association (2B)

2A: Pre-existing periodontitis exposure and incident diabetes outcome association

Cox model with Cox model with time-varying Crude Cox model baseline covariatesc covariatesd Person-years Exposure Incident Diabetes of follow-up IRR HR 95% CI HR 95% CI HR 95% CI

Periodontitisa Severe/moderate 45 4,536.45 1.40 1.46 (0.79-2.71) 1.21 (0.64-2.30) 1.33 (0.71-2.52) Mild/none 37 5,236.48 Ref Ref Ref Ref

2B: Pre-existing diabetes exposure and incident periodontitis outcome association

Cox model with Cox model with time-varying Crude Cox model baseline covariatesc covariatesd Person-years Exposure Incident Periodontitis of follow-up IRR HR 95% CI HR 95% CI HR 95% CI

Diabetesb Diabetic 7 92.68 1.54 1.45 (0.86-2.45) 1.35 (0.79-2.32) 1.32 (0.77-2.26) Non-diabetic 257 5,228.54 Ref Ref Ref Ref

Abbreviations: IRR, incidence rate ratio; HR, hazard ratio; CI, confidence interval; Ref, reference. aCox proportional hazards model of developing diabetes. bCox proportional hazards model of developing severe periodontitis. cCovariates were baseline age (continuous), baseline education (≥ College graduate and positive association while indicate <1 negative association). 58 | ALSHIHAYB et al. explain away the association between diabetes as an exposure and since the hazard ratios from the competing events models did not severe periodontitis as the outcome. In contrast, the simulations of differ from the models without the competing events by more than the impact of unmeasured confounding on the association between 10%, it was unlikely that participant death affected the simulations pre-existing periodontitis (exposure) and incident type II diabetes of unmeasured confounders. Moreover, tooth loss prior to the ob- (outcome) showed that the association could be partially or totally servation period could result in misspecification of periodontitis sta- explained away by the introduction of unmeasured confounding. tus and thus, potentially impact the original hazard ratio estimates Current understanding of the mechanisms driving the bidirec- and in turn, the quantitative bias analysis results. This is known as tional nature of the associations between periodontitis and diabetes prevalent cohort bias (Heaton et al., 2014). However, the effect was remains limited. Our results support the likelihood that the under- minimized by restricting eligibility to participants with 8 teeth at the lying causal mechanisms may differ in important ways, as the as- baseline exam. Additionally, in the association where pre-existing sociations clearly differ in the extent to which each is affected by periodontitis was the exposure, unobserved variables prior to study unknown confounding. Alternatively, a true bi-directionality of the baseline may have affected both the diabetes outcome and the peri- associations may not exist but, rather, its appearance results from odontitis status observed at study entry. Specifically, it is possible systematic error due to confounding, at least in the direction of that tooth loss prior to entry could play a confounding role in the ob- pre-existing periodontitis being causal to incident type II diabetes. served association between periodontitis and diabetes. However, in This study has some key strengths. Our longitudinal dataset per- the absence of causal knowledge, we cannot be sure whether tooth mitted us to evaluate each direction of the periodontitis-diabetes loss would confound the association or induce a bias by virtue of associations independently and to apply QBA to estimate how the its control. To investigate, we comprised a model that additionally sensitivities to unmeasured confounding might differ. The ability to controlled for the number of teeth at baseline (HR = 1.17; 95% CI: explore this bi-directionality in a single, well-characterized longitu- 0.61–2.23). The lack of a meaningful change in the HR leads us to dinal cohort is uncommon, and the use of QBA is new to periodontal believe that tooth loss prior to baseline did not play a major role in research. There are other, less comprehensive approaches to ex- the association under study. plore the effects of unmeasured confounding, namely the e-value Other limitations of this study included the potential bias from (VanderWeele & Ding, 2017), which identifies the conditions that differential selection or differential loss to follow-up. However, the are minimally necessary to bring the observed estimate to the null impact of this is expected to be minimal since there were only minor value. Our results in which positive confounding totally explained differences in loss to follow-up between exposure groups. The ap- the associations are comparable to the e-value. proach to ascertain diabetes and/or periodontitis, define their time This study also has some important limitations. QBA is an of occurrence and address artificial censoring of participants could emerging approach to understanding the potential role of unmea- each have introduced misclassification or selection biases in the sured confounding. Limitations that exist in the current state of QBA original analyses’ hazard ratios (i.e. analyses without the unmea- method development are reflected in this study. For example, we sured confounder). For example, the lack of clinical measures at simulated confounder associations based on the exposure status at baseline may have resulted in inclusion of participants with undiag- the time of censoring or failure. Thus, they were time-fixed and did nosed diabetes at study entry. Therefore, the QBA results may differ not vary in this analysis. This could have resulted in misclassified per- from what we observed as they depend on the underlying or “true” son-time of the simulated confounder at exams prior to the last one. hazard ratio. The lack of precision (e.g. random error) in the underly- However, this was a better option than simulating the confounders ing effect estimates is also a potential limitation. However, the QBA at baseline (as no person had the particular outcome of interest at relies on the point estimate, which we believe to have been mea- that time point). Doing so would also have resulted in misclassified sured validly. Thus, replication of our findings by applying QBA in person-time of the simulated confounder at later exams (i.e. after other cohorts may be a promising avenue for future work. Lastly, our baseline). Both of these could introduce residual confounding and cohort only included males who predominantly identified as White, it may not be known which option had the least residual confound- Non-Hispanic. Thus, our results may not be generalizable outside ing. Additionally, our QBA only introduced one unmeasured time- this population if true differences by gender or race/ethnicity are fixed confounder. If the hypothesized unmeasured confounded present for this association. is time-varying, the results may differ. In addition, there could be Our findings suggest that the bidirectional associations between several unmeasured confounders and simply controlling for one periodontitis and diabetes may not be truly bidirectional in regards would still result in residual confounding in both associations. QBA to causality. Rather, some observed associations may be artifactual methods have not yet been developed to account for such relevant due to an unmeasured or unknown common cause. Given that both scenarios. conditions result from inflammatory processes, it is plausible, for ex- In the analyses of pre-existing periodontitis as the exposure and ample, that each condition shares a common inflammatory cause, incident type II diabetes as the outcome, the competing event of that is, confounder, which is independently associated with both death could have modified the simulation of an unmeasured con- periodontitis and diabetes. A range of parameters reflecting this founder by affecting the proportion of participants expected to have scenario were explored in our QBA, the results of which suggest the confounder among the exposure and disease groups. However, that confounding may be a plausible explanation for the observed ALSHIHAYB et al. | 59 periodontitis-diabetes association. We also explored the possibility Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. (2003). Report of the expert committee on the diagno- of a common pathway that had opposite effects on the studied ex- sis and classification of diabetes mellitus. Diabetes Care, 26, S5-20. posure and outcome, that is, positively associated with one condi- https://doi.org/10.2337/diacare.26.2007.s5 tion, but negatively associated with the other. While this scenario Feldman, R. S., Douglass, C. W., Loftus, E. R., Kapur, K. K., & Chauncey, resulted in the presence of negative confounding, that is, bias to- H. H. (1982). Interexaminer agreement in the measurement of periodontal disease. , , 80-89. https:// wards the null, based on our current causal knowledge with respect Journal of Periodontal Research 17 doi.org/10.1111/j.1600-0765.1982.tb01133.x to periodontitis and diabetes, it would be premature to suggest what Greenland, S., Pearl, J., & Robins, J. M. (1999). Causal diagrams for epide- such a variable may be. A next step to better understand the causal miologic research. Epidemiology, 10, 37-48. bidirectional nature of the periodontitis-diabetes association could Heaton, B., Applebaum, K. M., Rothman, K. J., Brooks, D. R., Heeren, T., be achieved by mediation analysis. Dietrich, T., & Garcia, R. I. (2014). The influence of prevalent cohort bias in the association between periodontal disease progression QBA is a relatively simple tool that can be used to increase and incident coronary heart disease. Annals of Epidemiology, 24, the rigour of investigations into periodontitis-systemic disease 741-746. https://doi.org/10.1016/j.annepidem.2014.07.006 associations. In light of our still limited understanding of these as- Hujoel, P. P., Drangsholt, M., Spiekerman, C., & DeRouen, T. A. (2000). sociations, and their underlying causal mechanisms or pathways, Periodontal disease and coronary heart disease risk. JAMA, 284, 1406-1410. https://doi.org/10.1001/jama.284.11.1406 it is crucial to address the uncertainty of these observed associa- Joshipura, K. J., Douglass, C. W., & Willett, W. C. (1998). Possible ex- tions as being due to plausible sources of bias, including unknown planations for the tooth loss and cardiovascular disease relation- confounding. Quantification of this uncertainty about sources ship. Annals of Periodontology, 3, 175-183. https://doi.org/10.1902/ of bias through the use of quantitative bias analysis should be annals.1998.3.1.175 Kapur, K. K., Glass, R. L., Loftus, E. R., Alman, J. E., & Feller, R. P. highly informative to research on periodontitis-systemic disease (1972). The Veterans Administration longitudinal study of oral associations. health and disease: methodology and preliminary findings. Aging and Human Development, 3, 125-137. https://doi.org/10.2190/ CONFLICT OF INTEREST WLL4-ET76-UQWN-R5FL Khader, Y. S., Dauod, A. S., El-Qaderi, S. S., Alkafajei, A., & Batayha, W. Q. The authors have stated explicitly that there are no conflicts of inter- (2006). Periodontal status of diabetics compared with nondiabet- est in connection with this article. ics: A meta-analysis. Journal of Diabetes and Its Complications, 20, 59-68. https://doi.org/10.1016/j.jdiacomp.2005.05.006 ORCID Lash, T. L., Fox, M. P., & Fink, A. K. (2009). Unmeasured and unknown confounders. In T. L. Lash, M. P. Fox, & A. K. Fink (Eds.), Applying Talal S. Alshihayb https://orcid.org/0000-0001-5750-4144 quantitative bias analysis to epidemiologic data (pp. 58-78). Springer Brenda Heaton https://orcid.org/0000-0002-3097-4717 Science & Business Media. Lash, T. L., Fox, M. P., MacLehose, R. F., Maldonado, G., McCandless, L. C., REFERENCES & Greenland, S. (2014). Good practices for quantitative bias analysis. , , 1969-1985. https://doi. Aarabi, G., Zeller, T., Seedorf, H., Reissmann, D. R., Heydecke, G., International Journal of Epidemiology 43 org/10.1093/ije/dyu149 Schaefer, A. S., & Seedorf, U. (2017). Genetic susceptibility con- Leffondre, K., Abrahamowicz, M., Xiao, Y., & Siemiatycki, J. (2006). tributing to periodontal and cardiovascular disease. Journal of Modelling smoking history using a comprehensive smoking index: Dental Research, 96, 610-617. https://doi.org/10.1177/00220 Application to lung cancer. , , 4132-4146. 3 4 5 1 7 6 9 9 7 8 6 Statistics in Medicine 25 https://doi.org/10.1002/sim.2680 Akinkugbe, A., Sharma, S., Ohrbach, R., Slade, G., & Poole, C. (2016). Linden, G. J., Lyons, A., & Scannapieco, F. A. (2013). Periodontal systemic Directed acyclic graphs for oral disease research. Journal of Dental associations: Review of the evidence. Journal of Periodontology, 84, Research, 95, 853-859. https://doi.org/10.1177/0022034516 S8-S19. https://doi.org/10.1902/jop.2013.1340010 639920 Löe, H. (1993). Periodontal disease: The sixth complication of diabetes Allison, P. D. (2010). Estimating Cox regression models with PROC mellitus. Diabetes Care, 16, 329-334. https://doi.org/10.2337/diaca PHREG. In P. D. Allison (Ed.), Survival analysis using SAS: A practical re.16.1.329 guide (pp. 125-200). SAS Institute. Madianos, P. N., & Koromantzos, P. A. (2018). An update of the evidence Bell, B., Rose, C. L., & Damon, A. (1966). The Veterans Administration on the potential impact of periodontal therapy on diabetes out- longitudinal study of healthy aging. Gerontologist, 6, 179-184. comes. , , 188-195. https://doi. https://doi.org/10.1093/geront/6.4.179 Journal of Clinical Periodontology 45 org/10.1111/jcpe.12836 Chavarry, N. G. M., Vettore, M. V., Sansone, C., & Sheiham, A. (2009). The Mealey, B. L., & Ocampo, G. L. (2007). Diabetes mellitus and peri- relationship between diabetes mellitus and destructive periodontal odontal disease. Periodontology 2000, 44(1), 127-153. https://doi. disease: A meta-analysis. Oral Health & Preventive Dentistry, 7, 107- org/10.1111/j.1600-0757.2006.00193.x 127. https://doi.org/10.3290/j.ohpd.a15518 Merchant, A. T., & Pitiphat, W. (2002). Directed acyclic graphs (DAGs): Demmer, R. T., Holtfreter, B., Desvarieux, M., Jacobs, D. R., Kerner, W., an aid to assess confounding in dental research. Nauck, M., Volzke, H., & Kocher, T. (2012). The influence of type 1 Community , , 399-404. https://doi. and type 2 diabetes on periodontal disease progression: Prospective Dentistry and Oral Epidemiology 30 org/10.1034/j.1600-0528.2002.00008.x results from the Study of Health in Pomerania (SHIP). Diabetes Care, Page, R. C., & Eke, P. I. (2007). Case definitions for use in population 35(10), 2036-2042. https://doi.org/10.2337/dc11-2453 – Based surveillance of periodontitis. , , Dietrich, T., & Hoffmann, K. (2004). A comprehensive index for the mod- Journal of Periodontology 78 1387-1399. https://doi.org/10.1902/jop.2007.060264 eling of smoking history in periodontal research. Journal of Dental Polak, D., & Shapira, L. (2018). An update on the evidence for patho- Research, 83, 859-863. https://doi.org/10.1177/1544059104 genic mechanisms that may link periodontitis and diabetes. 08301107 Journal 60 | ALSHIHAYB et al.

of Clinical Periodontology, 45, 150-166. https://doi.org/10.1111/ Textor, J., Hardt, J., & Knuppel, S. (2011). DAGitty: a graphical tool for jcpe.12803 analyzing causal diagrams. Epidemiology, 22, 745. https://doi. Preshaw, P. M., Alba, A. L., Herrera, D., Jepsen, S., Konstantinidis, A., org/10.1097/EDE.0b013e318225c2be Makrilakis, K., & Taylor, R. (2012). Periodontitis and diabetes: a two- VanderWeele, T. J., & Ding, P. (2017). Sensitivity analysis in observational way relationship. Diabetologia, 55, 21-31. https://doi.org/10.1007/ research: Introducing the E-value. Annals of Internal Medicine, 167, s00125-011-2342-y 268-274. Salvi, G. E., Carollo-Bittel, B., & Lang, N. P. (2008). Effects of diabetes Winning, L., Patterson, C. C., Neville, C. E., Kee, F., & Linden, G. J. (2017). mellitus on periodontal and peri-implant conditions: update on as- Periodontitis and incident type 2 diabetes: A prospective cohort sociations and risks. Journal of Clinical Periodontology, 35, 398-409. study. Journal of Clinical Periodontology, 44(3), 266-274. https://doi. https://doi.org/10.1111/j.1600-051X.2008.01282.x org/10.1111/jcpe.12691 Sanz, M., Ceriello, A., Buysschaert, M., Chapple, I., Demmer, R. T., Graziani, F., Herrera, D., Jepsen, S., Lione, L., Madianos, P., Mathur, M., Montanya, E., Shapira, L., Tonetti, M., & Vegh, D. (2018). Scientific evidence on the links between periodontal diseases and diabetes: SUPPORTING INFORMATION Consensus report and guidelines of the joint workshop on periodon- Additional supporting information may be found online in the tal diseases and diabetes by the International Diabetes Federation Supporting Information section. and the European Federation of Periodontology. Journal of Clinical Periodontology, 45, 138-149. https://doi.org/10.1111/jcpe.12808 Simpson, T. C., Needleman, I., Wild, S. H., Moles, D. R., & Mills, E. J. How to cite this article: Alshihayb TS, Kaye EA, Zhao Y, (2010). Treatment of periodontal disease for glycaemic control in people with diabetes. Cochrane Database of Systematic Reviews, Leone CW, Heaton B. A quantitative bias analysis to assess CD004714. https://doi.org/10.1002/14651858.CD004714.pub2 the impact of unmeasured confounding on associations Simpson, T. C., Weldon, J. C., Worthington, H. V., Needleman, I., Wild, between diabetes and periodontitis. J Clin Periodontol. S. H., Moles, D. R., & Iheozor-Ejiofor, Z. (2015). Treatment of peri- 2020;48:51–60. https://doi.org/10.1111/jcpe.13386 odontal disease for glycaemic control in people with diabetes mellitus. Cochrane Database of Systematic Reviews, CD004714. https:// doi.org/10.1002/14651858.CD004714.pub3 Received: 31 March 2020 | Revised: 17 September 2020 | Accepted: 28 September 2020 DOI: 10.1111/jcpe.13384

SYSTEMATIC REVIEW

The impact of smoking on non-surgical periodontal therapy: A systematic review and meta-analysis

Jennifer Chang1 | Hsiu-Wan Meng1 | Evanthia Lalla2 | Chun-Teh Lee1

1Department of Periodontics and Dental Hygiene, The University of Texas Health Abstract Science Center at Houston School of Aim: Smoking is a risk factor for periodontitis. This study aimed to evaluate the impact Dentistry, Houston, TX, USA 2Division of Periodontics, Columbia of smoking on clinical outcomes of non-surgical periodontal therapy. University College of Dental Medicine, Materials and Methods: Electronic databases were searched to screen studies pub- New York, NY, USA lished before May 2020. The included studies had to have two groups: smokers (S) Correspondence and non-smokers (NS) with periodontitis. The outcomes evaluated were differences Chun-Teh Lee, Department of Periodontics and Dental Hygiene, The University of between groups in probing depth (PD) reduction and clinical attachment level (CAL) Texas Health Science Center at Houston gain after non-surgical periodontal therapy. Meta-regressions were conducted to School of Dentistry, 7500 Cambridge Street, Suite 6470, Houston, TX, USA. evaluate correlations between outcomes and other contributing factors. Email: [email protected] Results: Seventeen studies were included. The post-treatment PD reduction in the S group was smaller than in the NS group (weighted mean difference in PD reduction: −0.33 mm, 95% confidence interval (CI): [−0.49, −0.17], p < .01). The CAL gain in the S group was also smaller than in the NS group (weighted mean difference in CAL gain: −0.20 mm, CI: [−0.39, −0.02], p < .01). Additionally, baseline PD significantly affected the difference in PD reduction between two groups. Conclusions: Smoking negatively impacts clinical responses to non-surgical periodontal therapy. Smokers with periodontitis have significantly less PD reduction and CAL gain than non-smokers.

KEYWORDS cigarette smoking, dental scaling, evidence-based dentistry, root planning, treatment outcome

1 | INTRODUCTION loss and more missing teeth compared to former or never smokers. Such results are in agreement with several other studies indicat- Smoking has been identified as a major risk factor for periodontal ing that smokers have more severe periodontal destruction than diseases (Genco & Borgnakke, 2013). Previous longitudinal studies non-smokers (Axelsson et al., 1998; Baljoon et al., 2005; Gunsolley have found that smokers have accelerated and more severe peri- et al., 1998; Montero et al., 2019). The negative effect of nicotine in odontal tissue destruction, poor wound healing, and respond less tobacco on cells is confirmed by in vitro studies (Ryder et al., 1998; favourably towards periodontal treatments (Feldman et al., 1987; Van Der Weijden et al., 2001). Helal et al., 2019; Krall et al., 1999; Preber & Bergström, 1990). Regarding treatment outcomes following periodontal ther- A cross-sectional study (Haffajee & Socransky, 2001) reported apy, a prospective split-mouth longitudinal study (Ah et al., 1994) that smokers had significantly deeper pockets, greater attachment demonstrated that smokers had significantly less probing depth (PD)

wileyonlinelibrary.com/journal/jcpe | 61 J Clin Periodontol. 2021;48:61–76. 62 | CHANG et al. reduction and clinical attachment level (CAL) gain after non-surgical and surgical periodontal therapy. Moreover, there was greater at- Clinical Relevance tachment loss in smokers during maintenance. Scientific rationale for the study: Smoking has been identi- Although several studies agreed upon worse periodontal fied as a major risk factor for periodontitis. However, there disease and less favourable response to periodontal therapy in is no systematic review and meta-analysis comparing the smokers (Boström et al., 1998; Heasman et al., 2006; Preber & changes of PD and CAL following non-surgical periodontal Bergström, 1990; Tonetti et al., 1995), some studies suggested therapy in smokers and non-smokers. otherwise (D’Aiuto et al., 2005; Guru et al., 2018; Preshaw et al., Principal findings: Smokers with periodontitis had signifi- 2013). A prospective cohort study (Guru et al., 2018) revealed no cantly less PD reduction and CAL gain than non-smokers significant differences between smokers and non-smokers in PD with periodontitis within one year following non-surgical (respective mean of 1.96 and 1.98 mm) and CAL (respective mean periodontal therapy (both p < .01). of 1.32 and 1.29 mm) changes one month after full-mouth scaling Practical implications: Clinicians should expect inferior clin- and root planning. Additionally, several other studies did not find ical outcomes of non-surgical periodontal therapy in smok- significant differences between smokers and non-smokers regard- ers compared to non-smokers. ing outcomes of surgical treatments (Machtei et al., 1998), including soft tissue grafting (Harris, 1994) and dental implant placement (Weyant, 1994). With such controversies from previous studies, there is a critical 4. Studies reporting the outcomes of PD and/or CAL at baseline and need to comprehensively evaluate the impact of smoking on clinical at one-month to one-year follow-up, and/or changes in PD and outcomes following periodontal treatments. Therefore, this system- CAL following treatment. atic review and meta-analysis aimed to compare the PD reduction and CAL gain following non-surgical periodontal therapy in smokers versus non-smokers. 2.3 | Search Strategy

A search was conducted for studies published from January 1960 to 2 | MATERIALS AND METHODS May 2020 in four electronic databases, MEDLINE (PubMed), Embase, Scopus and ClinicalTrials.gov, using specific key terms (Appendix This systematic review was performed following the Preferred S1). Additionally, a search was performed in the following journals: Reporting Items for Systematic Reviews and Meta-Analyses Journal of Periodontology, Journal of Clinical Periodontology, Journal of (PRISMA) guidelines (Liberati et al., 2009; Moher et al., 2015). Periodontal Research and Journal of Dental Research. Articles in the bibliography of the screened studies were also reviewed if these articles might fulfil the selection criteria. 2.1 | Focused Question

Based on the PICO principle—the Population: smokers with peri- 2.4 | Quality Assessment odontitis (S); the Intervention: non-surgical periodontal therapy; the Control: non-smokers with periodontitis (NS); the Outcomes of The Newcastle–Ottawa scale (NOS) was used to assess the methodo- interest: PD reduction and CAL gain after non-surgical periodontal logical quality, including the selection (score range 0–4), the compa- therapy. rability (score range 0–2) and the outcome (score range 0–3) of the The following focused question was proposed: “What is the im- selected studies (Wells et al., 2000). Therefore, the total NOS score pact of smoking on PD reduction and CAL gain following non-surgi- for each study can range from 0 to 9, with 9 being the highest possible cal periodontal therapy in smokers with periodontitis compared to score. The higher the score, the better the methodological quality. non-smokers with periodontitis during the first year of follow-up?”. Additionally, the evidence level of each study was evaluated following the Oxford Centre for Evidence-based Medicine recommendation (Phillips et al., 2011). The quality assessment was conducted 2.2 | Study Selection Criteria by two authors (J.C. and H.M.) independently, and the agreement between the two authors was evaluated using the kappa statistic. 1. Randomized controlled trials, case-control studies, prospective or retrospective cohort studies. 2. Studies including at least two groups: smokers with periodontitis 2.5 | Grading the Body of Evidence (S) and non-smokers with periodontitis (NS). 3. Studies including patients receiving only non-surgical periodontal Quality-of-evidence (risk of bias in reported outcomes, inconsistency therapy. of outcomes among studies, indirectness of reported outcomes, CHANG et al. | 63 imprecision of reported outcomes and potential publication bias) the known t value (Follmann et al., 1992; Higgins & Green, 2011; Lee assessment was done based on the Grading of Recommendations et al., 2016). Assessment, Development and Evaluation (GRADE) system (Guyatt Heterogeneity of data sets for the outcomes of interest was et al., 2008, 2011). The quality of the studies was independently tested using chi-squared test of homogeneity. If the heterogeneity evaluated by two authors (J.C. and H.M.). If needed, agreement on was significant (p < .05), random effects model (DerSimonian–Laird evidence level was reached by discussion (Appendix S2). test) was chosen to perform the meta-analysis. If the heterogeneity was not significant (p ≥ .05), fixed effects model was used (Higgins & Green, 2011). Heterogeneity between data sets was also 2.6 | Data Extraction and Data Synthesis assessed using I2 statistic describing the variation of each data set (I2 < 30%: low heterogeneity; I2 = 30%–60%: moderate heteroge- The titles and abstracts of all retrieved articles were independently neity; I2 > 60%: considerable heterogeneity; Alqaderi et al., 2016; screened by two authors (J.C. and H.M.). Unrelated articles were Higgins & Green, 2011). Forest plots were generated to demon- excluded. The full texts of potentially qualifying articles were com- strate the individual and pooled effect estimates, as well as the prehensively reviewed. Disagreement between the two authors 95% confidence intervals (CI). Meta-regressions were conducted to was planned to solve by discussion when necessary. If the identified evaluate the correlation between the outcomes of interest (PD and studies had multiple groups of patients, only the groups meeting the CAL changes) and different variables. To evaluate if the variables selection criteria described above were included. If outcomes at mul- caused heterogeneity or bias, subgroup analyses were conducted. tiple time points were reported, only the outcomes with the longest Sensitivity analyses were performed to assess the robustness of follow-up within 1 year post-treatment were included. Data were the meta-analysis results by omitting one data set in each step or independently extracted by two authors (J.C. and H.M.) with a spe- omitting specific data sets. Publication bias was evaluated by Egger's cially designed data extraction form, and the accuracy of extracted tests and funnel plots. All statistical analyses were performed using data was confirmed by a third author (C.L.). The authors of the se- STATA® (Version 15.1, 2017, Stata Corp). Statistical significance was lected articles were contacted if there was unclear information. defined as p < .05.

2.7 | Data Analysis 3 | RESULTS

The primary outcome was the difference in PD reduction after non- 3.1 | Study selection surgical therapy between the smokers with periodontitis group (S) and the non-smokers with periodontitis group (NS). PD was defined Two thousand three hundred and forty-two articles were identified as the distance from the gingival margin to the tip of the periodontal using the search strategies (Appendix S1). After screening the titles probe when probing. PD reduction was calculated by subtracting the and abstracts, 41 articles potentially qualified. Twenty-four articles PD at the follow-up visit from the PD at baseline. Weighted mean were excluded because they did not report the sample size or the difference in PD reduction (WDPD) was calculated by subtracting mean values or standard deviations in PD or CAL before or after non- the mean PD reduction in NS from the mean PD reduction in S. The surgical periodontal therapy, only reported measurement changes difference in PD reduction between the two groups in each study of deep probing sites, included additional treatment procedures, or was pooled based on the assigned weights. A positive value means reported follow-up results not within the one-month to one-year the S had more PD reduction than the NS, and a negative value range (Appendix S3; Abreu et al., 2019; Ah et al., 1994; Camargo means the S had less PD reduction than the NS. et al., 2016; Chandra et al., 2012; Chang et al., 2018; Christan et al., The secondary outcome was the difference in CAL after non-sur- 2002; Darby et al., 2005; Erdemir et al., 2004; Farina et al., 2010; gical therapy between S and NS. CAL was defined as the distance Gleissner et al., 2003; Grossi et al., 1997; Hendek et al., 2015; Jiao from the cemento-enamel junction to the tip of a periodontal probe et al., 2017; Kaldahl et al., 1996; Meulman et al., 2012; Mohan et al., during periodontal probing. CAL gain was calculated by subtracting 2019; Nassrawin, 2010; Pamuk et al., 2017; Preber & Bergström, the CAL at the follow-up visit from the CAL at baseline. Weighted 1986b; Preshaw et al., 1999; Preus et al., 2014; Pucher et al., 1997; mean difference in CAL gain (WDCAL) was calculated by subtracting Zuabi et al., 1999). Seventeen studies were finally selected, and 948 the mean CAL gain in NS from the mean CAL gain in S. The differ- patients were included in these selected studies (Table 1). ence in CAL gain between the two groups in each study was pooled based on the assigned weights. A positive value means the S group had more CAL gain than the NS group, and a negative value means 3.2 | Study characteristics the S group had less CAL gain than the NS group. Some of the included studies did not provide SDs of the changes Eleven selected studies were prospective cohort studies, and six in PD or CAL after treatment. Such missing SDs were imputed by selected studies were retrospective cohort studies. Most selected using an assumptive correlation coefficient which equals to 0.5 or studies defined the number of cigarettes smoked in smokers, but 64 | CHANG et al.

TABLE 1 Overview of the selected studies

Demographic data

Author (year) Groups Subject number Age (year) Gender (Female %)

AlAhmari et al. (2019) S 22 45.2 ± 3.6 0 NS 21 44.2 ± 2.4 0 Guru et al. (2018) S 23 36.4 ± 8.2 NA NS 23 36.3 ± 8.4 NA Türkoğlu et al. (2016) S 16 43.5 ± 7.8 44 NS 15 46.4 ± 7.2 47 Dodington et al. (2015) S 23 53 ± 7 57 NS 63 59 ± 12 48 Feres et al. (2015) S 15 40.5 ± 8.2 47 NS 15 42.1 ± 6.5 47 Ardais et al. (2014) S 11 47 ± 7.81 64 NS 14 47.64 ± 10.81 64 Preshaw et al. (2013) S 40 47.6 ± 9.1 55 NS 155 54.8 ± 11.5 57 Eltas and Orbak, (2012) S 26 NA 50 NS 26 NA 50 Wan et al. (2009) S 20 46.2 ± 6.8 0 NS 20 45 ± 5.9 0 Hughes et al. (2006) S 20 35.3 ± 4.19 NA NS 59 33.68 ± 5.33 NA Apatzidou et al. (2005) S 15 41 40 NS 25 45 44 D’Aiuto et al. (2005) S 55 NA NA NS 39 NA NA Jin et al. (2000) S 13 NA NA NS 12 NA NA Palmer et al. (1999) S 9 50 NA NS 18 50.7 NA Renvert et al. (1998) S 13 43 NA NS 15 45 NA Preber et al. (1995) S 17 48.5 76 NS 15 49.1 53 Preber & Bergström (1986a) S 40 39.5 57.5 NS 35 41.8 63 Periodontal management

Follow-up Subject oral hygiene, Author (year) Periodontal status Smoking status Non-surgical treatment (months) motivation and compliance

AlAhmari et al. CP (at least 30% sites with 11.7 ± 0.3 pack-years SRP under local 1 and 3 73.8% smokers and 80.5% (2019) PD ≥3 mm and CAL anaesthesia, using non-smokers reported ≥3 mm) hand instrument, tooth brushing twice daily by a trained dental at 3-month follow-up hygienist, within one session Guru et al. Generalized moderate to ≥20 cigarettes/day for SRP, using ultrasonic and 1 Two smokers and two non- (2018) severe CP (CAL ≥3 mm) >2 years hand instruments smokers did not return for A minimum of 20 permanent 30-day evaluation teeth CHANG et al. | 65

Follow-up Subject oral hygiene, Author (year) Periodontal status Smoking status Non-surgical treatment (months) motivation and compliance

Türkoğlu et al. Moderate to Severe CP (at ≥10 cigarettes/day for SRP under local 1 and 3 Periodontal treatment (2016) least 30% bone loss, ≥10 years anaesthesia, using included motivation and ≥4 non-adjacent teeth ultrasonic and hand oral hygiene instructions. with PD ≥5 mm and instruments, by two Two smokers and one CAL≥4 mm) trained periodontists, non-smoker discontinued. within 14 days Two non-smokers who did (number of visits not perform proper plaque determined based on control were excluded patient's requisites) Dodington Generalized CP (at least 30% Current smokers SRP, using ultrasonic and 2–4 Oral hygiene instructions were et al. sites with PD ≥4 mm) instruments, by four given. (2015) hygienists, within one 17 were lost to follow-up, and session 11 did not follow up within 4 months Feres et al. Moderate to severe CP (had ≥10 cigarettes/day SRP, using hand 6 All subjects were monitored to (2015) at least 15 teeth, ≥6 teeth within the past instruments, by a ensure good oral hygiene with at least one site 5 years (17.1 ± 7.3 periodontist, within practices during the of PD 5–7 mm and CAL cigarettes/day for six visits (1 h each in experiment. 5–10 mm) 26.6 ± 7.3 years) 21 days) No dropouts. Ardais et al. Periodontitis (≥4 teeth with ≥10 cigarettes/day for SRP under local 3 Oral hygiene instruction and (2014) PD≥5 mm, CAL≥5 mm, ≥6 months anaesthesia, by professional removal of and BOP one or more (19.27 ± 3.28 an experienced supragingival biofilm were sites) cigarettes/day for periodontist done every 2 weeks during 25.68 ± 9.52 years) the experiment. Two smokers and one non- smoker lost follow-up Preshaw et al. CP (≥6 teeth with PD≥5 mm) 14.3 ± 6.2 cigarettes/day SRP under local 3 Individually tailored oral (2013) for 18 ± 5.4 years anaesthesia, using hygiene instruction and ultrasonic and hand motivation were given. instruments, by 29 patients did not complete hygienists in training, the course of treatment within 6–8 sessions Eltas and Generalized moderate CP (≥2 >10 cigarettes/day for SRP under local 1 and 6 Thorough oral hygiene Orbak teeth with PD 4-6 mm >5 years (17 ± 6 anaesthesia, using instructions were given. (2012) and radiographic bone cigarettes/day) ultrasonic and hand All patients completed the loss per quadrants) instruments, within 6-month evaluation one session Wan et al. CP (≥16 remaining teeth ≥10 cigarettes/day for Quadrant-wise 3, 6 and 12 Oral hygiene instructions (2009) with at least one tooth ≥10 years (20.8 ± 8.7 debridement under regarding brushing and having PD≥5 mm in each pack-years) local anaesthesia, inter-dental cleaning were quadrant, excluding third by six experienced given. molars) hygienists, within 3 smokers and 2 non-smokers eight weeks (four to lost follow-up six visits) Hughes et al. Generalized aggressive Self-reported current Full-mouth subgingival 2.5 Oral hygiene instructions were (2006) periodontitis (under age smokers or scaling and given 40, CAL ≥6 mm affecting smokerlyzer reading debridement using ≥6 teeth: >2 teeth in of >7 (CO>10p.p.m.) hand and ultrasonic addition to incisors and instrumentation first molars) one quadrant at a time under local anaesthesia by two experienced periodontists, over four visits of 30 min duration each at weekly intervals 66 | CHANG et al.

Follow-up Subject oral hygiene, Author (year) Periodontal status Smoking status Non-surgical treatment (months) motivation and compliance

Apatzidou Moderate to severe CP (≥2 Self-reported smokers SRP under local 6 Oral hygiene instructions et al. non-adjacent sites per confirmed by the anaesthesia, using were given and reinforced (2005) quad with PD≥5 mm and cotinine enzyme ultrasonic and hand during the experiment radiographic bone loss) inhibition assay instruments, within ≥5 cigarettes/day one session or by quadrants at two- weekly intervals D’Aiuto et al. Generalized severe CP (PD Self-reported current SRP under local 6 Oral hygiene instructions were (2005) ≥5 mm with BOP and smokers anaesthesia, by given marginal alveolar bone an experienced loss >30% in at least 50% periodontist, using of the dentition) ultrasonic instrument, within one to three months (no limitation in time and visits) Jin et al. (2000) Severe CP (PD≥5 mm, >15 cigarettes/day for Scaling and root 1, 3 and 6 Oral hygiene instructions CAL≥3 mm and >10 years debridement, by a were given and reinforced radiographic bone loss on senior hygienist, in a at every visit during the at least 2 teeth/quad) course of total four to 6-week treatment period five hours Palmer et al. Moderate to severe Current smokers Subgingival scaling under 2 and 6 Oral hygiene instructions were (1999) periodontitis local anaesthesia, given (use of disclosing by a hygienist, tablets, toothbrush, using ultrasonic floss and interproximal instruments, within brushes). Oral hygiene two appointments reinforcement and (90 minutes each and supragingival prophylaxis 1 week apart) were carried out 1 and 4 weeks after SRP. Renvert et al. Severe periodontitis (≥3 sites ≥15 cigarettes/day SRP under local 6 Oral hygiene instructions (1998) with PD ≥6 mm and BOP) anaesthesia, by a were given and reinforced hygienist 3 months after SRP with supragingival rubber cup polishing Preber et al. Moderate to severe 21.8 cigarettes/day for Subgingival debridement, 2 Oral hygiene instructions were (1995) periodontitis (≥4 sites 23.6 years by a hygienist, given with PD≥6 mm and within 6 months (6-8 radiographic bone loss) sessions) Preber & Moderate to severe ≥20 cigarettes/day for SRP, by a hygienist, using 1 Oral hygiene instruction and Bergström periodontitis (PD≥4 mm) ≥5 years hand instruments, motivation were given and (1986a) within 5 months reinforced (average 7.8 1-h Patients were instructed in sessions per subject) tooth brushing and use of inter-dental cleaning aids. At each treatment session, oral hygiene was checked with disclosing solution and re- instructed when needed

Clinical parameters

PD (mm) CAL (mm) BOP (%)

Author (year) Groups Baseline Final Difference Baseline Final Difference Baseline Final Difference

Clinical parameters

AlAhmari et al. S 6.10 ± 0.50 5.50 ± 1.20 0.6 ± 1.04 7.10 ± 0.70 6.00 ± 0.40 1.10 ± 0.61 35.30 ± 6.90 32.70 ± 2.40 2.60 ± 6.07

(2019) NS 6.60 ± 0.40 4.10 ± 0.50 2.5 ± 0.46 7.20 ± 0.50 4.90 ± 0.60 2.30 ± 0.56 61.20 ± 9.60 20.30 ± 4.70 40.90 ± 8.31

Guru et al. S 4.95 ± 0.65 3.10 ± 0.75 1.85 ± 0.71 4.87 ± 0.24 3.55 ± 0.57 1.32 ± 0.50 NA NA NA

(2018) NS 4.31 ± 0.82 2.36 ± 0.42 1.95 ± 0.71 4.14 ± 0.35 2.91 ± 0.79 1.23 ± 0.69 NA NA NA CHANG et al. | 67

PD (mm) CAL (mm) BOP (%)

Author (year) Groups Baseline Final Difference Baseline Final Difference Baseline Final Difference

Türkoğlu et al. S 3.76 ± 0.51 2.61 ± 0.29 1.15 ± 0.44 4.87 ± 0.58 3.92 ± 0.67 0.95 ± 0.63 61.00 ± 22.00 7.90 ± 4.10 53.10

(2016) NS 3.53 ± 0.62 2.17 ± 0.28 1.36 ± 0.54 4.50 ± 0.98 3.56 ± 0.72 0.94 ± 0.88 88.00 ± 16.00 9.70 ± 5.50 78.30

Dodington S 4.26 ± 0.62 3.21 ± 0.20 1.05 ± 0.47 NA NA NA 48.00 ± 33.00 5.00 ± 8.00 43.00

et al. NS 4.08 ± 0.58 3.13 ± 0.11 0.95 ± 0.51 NA NA NA 50.00 ± 29.00 4.00 ± 1.00 46.00 (2015)

Feres et al. S 3.90 ± 0.60 3.30 ± 0.60 0.60 ± 0.60 4.70 ± 1.20 4.20 ± 0.90 0.50 ± 1.08 67.00 ± 18.20 53.90 ± 24.60 13.10

(2015) NS 3.80 ± 0.70 3.10 ± 0.40 0.70 ± 0.61 4.40 ± 1.00 3.90 ± 0.90 0.50 ± 0.95 62.00 ± 19.30 23.60 ± 14.10 38.40

Ardais et al. S 3.46 ± 0.47 2.69 ± 0.33 0.77 ± 0.36 4.94 ± 1.53 4.69 ± 1.58 0.24 ± 0.33 66.48 ± 19.02 50.00 ± 18.91 16.47 ± 15.25

(2014) NS 3.26 ± 0.44 2.37 ± 0.17 0.88 ± 0.37 2.95 ± 1.01 2.53 ± 0.97 0.41 ± 0.31 70.89 ± 13.00 47.07 ± 16.05 23.79 ± 12.93

Preshaw et al. S 3.60 ± 10 3.20 ± 0.90 0.40 ± 0.60 NA NA NA 41.50 ± 26.70 23.30 ± 20.10 18.20 ± 21.50

(2013) NS 3.10 ± 0.70 2.70 ± 0.60 0.40 ± 0.50 NA NA NA 36.50 ± 23.90 17.60 ± 17.50 19.70 ± 20.70

Eltas and S 5.40 ± 0.70 4.20 ± 0.50 1.2 ± 0.62 6.60 ± 1.20 6.50 ± 1.10 0.10 ± 1.15 NA NA NA

Orbak NS 5.30 ± 0.80 3.60 ± 0.50 1.7 ± 0.70 6.10 ± 0.90 5.80 ± 1.00 0.30 ± 0.95 NA NA NA (2012)

Wan et al. S 2.89 ± 0.52 2.01 ± 0.38 0.89 ± 0.32 3.71 ± 0.68 NA 0.31 ± 0.42 54.32 ± 13.68 26.91 ± 10.85 27.41

(2009) NS 2.82 ± 0.73 1.71 ± 0.28 1.11 ± 0.69 3.69 ± 0.97 NA 0.50 ± 0.52 73.45 ± 21.02 24.92 ± 10.44 48.53

Hughes et al. S NA NA 1.75 ± 0.56 NA NA 1.67 ± 0.73 43.00 ± 33.00 18.00 ± 20.00 25.00

(2006) NS NA NA 2.23 ± 0.87 NA NA 1.99 ± 0.74 61.00 ± 27.00 24.00 ± 17.00 37.00

Apatzidou S 4.40 ± 0.70 2.70 ± 0.30 1.70 ± 0.50 5.30 ± 1.10 4.20 ± 1.10 1.10 ± 0.50 67.30 ± 16.60 12.10 ± 7.90 55.20 ± 16.70

et al. NS 4.40 ± 0.60 2.50 ± 0.30 1.90 ± 0.60 4.90 ± 0.80 3.70 ± 0.80 1.20 ± 0.50 70.60 ± 18.70 11.60 ± 5.80 59.00 ± 19.80 (2005)

D’Aiuto et al. S 4.47 ± 0.82 3.43 ± 0.67 1.04 ± 0.45 NA NA NA NA NA NA

(2005) NS 4.30 ± 0.44 3.03 ± 0.31 1.26 ± 0.44 NA NA NA NA NA NA

Jin et al. S 4.60 ± 0.20 NA 1.10 ± 0.20 NA NA 0.40 ± 0.10 74.40 ± 6.10 40.10 ± 4.60 34.30

(2000) NS 4.50 ± 0.20 NA 1.60 ± 0.20 NA NA 0.60 ± 0.20 67.30 ± 5.00 31.20 ± 5.20 36.10

Palmer et al. S 5.73 ± 0.51 4.61 ± 0.70 1.12 ± 0.51 NA NA 0.47 ± 0.46 21.40 ± 9.40 12.10 ± 6.90 47.40 ± 12.60

(1999) NS 5.92 ± 0.49 3.96 ± 0.37 1.98 ± 0.44 NA NA 0.53 ± 0.43 24.20 ± 18.70 18.20 ± 16.30 28.20 ± 29.60

Renvert et al. S 7.10 ± 0.80 5.2 ± 0.90 1.90 ± 0.60 11.70 ± 1.10 10.70 ± 1.20 1.10 ± 0.90 63.00 ± 21.30 36.50 ± 19.90 26.50 ± 13.70

(1998) NS 7.60 ± 1.10 5.1 ± 0.80 2.50 ± 0.50 12.10 ± 1.30 11.10 ± 1.30 0.90 ± 0.60 53.00 ± 24.30 22.70 ± 12.30 30.90 ± 24.30

Preber et al. S NA NA 0.90 ± 0.30 NA NA NA NA NA NA

(1995) NS NA NA 1.10 ± 0.60 NA NA NA NA NA NA

Preber & S 4.50 ± 0.24 3.37 ± 0.36 1.13 ± 0.32 NA NA NA NA NA NA

Bergström NS 4.45 ± 0.20 3.22 ± 0.31 1.23 ± 0.27 NA NA NA NA NA NA (1986a)

Data are shown as mean ± SD if SD is available. S: smokers with periodontitis; N: non-smokers with periodontitis; SRP: scaling and root planing; CP: chronic periodontitis; PD: probing depth; CAL: clinical attachment level; BOP: bleeding on probing; and NA: not available. For studies reporting clinical outcomes more than once, only outcomes from the last follow-up within 1 year were analysed and presented in this table. two studies did not (Table 1). All the selected studies reported 3.3 | Quality assessments and mean PD reduction, and 12 of the 17 studies reported CAL gain heterogeneity evaluation after non-surgical periodontal therapy. Mean PD or CAL was calculated by the sum measurements (PD or CAL) of all sites divided The included studies had NOS scores ranging from 7 to 9, and the by the number of sites. All patients had periodontitis, but the ex- mean score was 8.53 ± 0.70 (Appendix S4). The evidence level of tent and severity of the disease were not consistent. According all the selected studies based on the Oxford Centre for Evidence- to the 2017 classification of periodontal diseases, the subjects based Medicine recommendation was 3. The kappa coefficients of the included articles had at least localized or generalized stage for the agreement in these two assessments between the two au- II grade A (grade B in smokers) periodontitis (Papapanou et al., thors (J.C. and H.M.) were both 1. The data sets for both WDPD 2018). and WDCAL had considerable heterogeneity (heterogeneity 68 | CHANG et al.

FIGURE 1 Pooled differences in probing depth (PD) reduction between the smokers with periodontitis (S) group and the non-smokers with periodontitis (NS) group. WMD: weighted mean difference

chi-squared p < .01 for both; I2 = 82.9 and 71.1% respectively; 3.5 | Meta-Regression Analyses Higgins & Green, 2011). Therefore, the random effects model (DerSimonian–Laird test) was used to perform the meta-analyses Variables, including follow-up period (months), age, smoking inten- for both WDPD and WDCAL. sity (heavy smoking defined as ≥20 cigarettes/day), proportion of sites with bleeding on probing (0.00–1.00), mean PD difference at baseline and mean CAL difference at baseline, were evaluated in the 3.4 | Synthesis of Results meta-regression analyses. Mean PD difference at baseline was directly correlated with the PD reduction difference between the two The differences in PD reduction and CAL gain following non-sur- groups (coefficient: 1.13, p = .02). Such result indicated that which- gical therapy between the S group and the NS group were pooled, ever group had the higher mean baseline PD would also have the and the estimates were calculated. The PD reduction in the S group greater PD reduction after non-surgical therapy (Table 2). None of was significantly lower than in the NS group (WDPD: −0.33 mm (CI: the variables were significantly related to the difference in CAL gain [−0.49, −0.17]), p < .01; Figure 1). The CAL gain in the S group was between the groups (Table 2). also significantly lower than in the NS group (WDCAL: −0.20 mm (CI: [−0.39, −0.02]), p < .01; Figure 2). Four studies provided data on PD reduction in deep pocket sites 3.6 | Subgroup Analyses and Sensitivity Analyses (≥5 mm; Apatzidou et al., 2005; Preber et al., 1995; Preshaw et al., 2013; Wan et al., 2009). The results showed that, in deep pock- Since the difference in mean baseline PD was significantly associ- ets, PD reduction in S was significantly lower than in NS. (WDPD: ated with PD reduction difference after non-surgical treatment be- −0.50 mm (CI: [−0.73, −0.26]), p < .01). tween the groups, the included articles were sub-grouped into (a) CHANG et al. | 69

FIGURE 2 Pooled differences in clinical attachment level (CAL) gain between the smokers with periodontitis (S) group and the non- smokers with periodontitis (NS) group. WMD: weighted mean difference mean baseline PD difference <0 (S < NS in mean baseline PD); (b) set of AlAhmari et al. (2019) was omitted (−0.15, CI: [−0.24, 0.07]). mean baseline PD difference ≥0 (S > NS in mean baseline PD); and (c) When the two studies, Guru et al. (2018) and Preber et al. (1995), mean baseline PD not reported. The mean baseline PD difference <0 with the shortest follow-up (1 month) were excluded, both PD re- subgroup appeared to favour PD reduction in NS (−1.11 [CI: −1.84, duction (WDPD: −0.37, CI [−0.56, −0.19]) and CAL gain (WDCAL: −0.38]) more than the mean baseline PD difference ≥0 subgroup −0.23, CI [−0.42, −0.04]) in the S were still significantly lower than (−0.18 [CI: −0.30, −0.06]) (Figure 3). in the NS. None of the data sets appeared to significantly affect the Additionally, the included studies were further sub-grouped pooled PD or CAL differences between two groups (Appendix S5). into (a) follow-up ≤3 months and (b) follow-up >3 months, due to the variation of follow-up lengths. The study by Dodington et al. (2015) reported the follow-up period ranged from 2 to 4 months 3.7 | Publication bias and was assigned to subgroup 1. Both subgroups showed PD reduction in S was significantly lower than NS (subgroup 1 WDPD: The results from Egger's test showed no significance in WDPD and −0.29 mm (CI: [−0.54, −0.04]), p = .02; subgroup 2 WDPD: −0.39 mm WDCAL (p = .14 and p = .86, respectively). The funnel plots generally (CI: [−0.55, −0.23]), p < .01; Figure 4). However, only the subgroup demonstrated symmetric distribution of the data sets (Appendices with >3 months of follow-up exhibited that the CAL reduction in S S6 and S7). Due to the limited number of included studies, publica- was significantly lower than NS (subgroup 1 WDCAL: −0.33 mm [CI: tion bias could not be completely ruled out. −0.78, 0.13], p = .16; subgroup 2: WDCAL: −0.16 mm (CI: [−0.26, −0.06]), p < .01). In the sensitivity analyses assessing the impact of data set from 3.8 | Quality of Evidence each article, WDPD was the lowest when Dodington et al. (2015) was omitted (−0.36 [CI: −0.53, −0.20]). WDPD was the highest Based on GRADE guidelines (Guyatt et al., 2008), the current evi- when AlAhmari et al. (2019) was omitted (−0.25, CI [−0.37, −0.13]). dence of the effect of smoking on PD reduction or CAL gain after WDCAL was the lowest when Guru et al. (2018) were omitted non-surgical periodontal treatment in periodontitis patients had (−0.23, CI: [−0.42, −0.04]). WDCAL was the highest when the data moderate quality (Appendix S4). 70 | CHANG et al.

TABLE 2 Association between patient-related variables and weighted mean difference in probing depth reduction (WDPD) (a) and clinical attachment level gain (WDCAL) (b) between groups.

p- Parameters Variables Coefficient (95% confidence interval) value

(a) WDPD Follow-up period −0.01 (−0.10, 0.08) .75 Age −0.09 (−0.19, 0.01) .08 Smoking intensity 0.14 (−0.16, 0.44) .32 Difference in proportion of sites with bleeding on 0.01 (−0.01, 0.04) .25 probing Mean probing depth difference at baseline 1.13 (0.48, 1.79) .02* (b) WDCAL Follow-up period 0.02 (−0.07, 0.10) .69 Age −0.10 (−0.26, 0.06) .19 Smoking intensity 0.26 (−0.18, 0.69) .21 Difference in proportion of sites with bleeding on 0.02 (−0.003, 0.04) .09 probing Mean clinical attachment level difference at baseline 0.10 (−0.43, 0.63) .67

Follow-up period: the time period in months between the baseline measurement and the last post-treatment measurement. Age: mean age difference of patients between the group of smokers with periodontitis (S) and the group of non-smokers with periodontitis (NS). Smoking intensity: heavy smokers (≥20 cigarettes per day) or not. Difference in proportion of sites with bleeding on probing: the difference of bleeding on probing (0.00–1.00) between the S group and the NS group at baseline. Mean probing depth difference at baseline: the difference of mean probing depth between the S group and the NS group at baseline. Mean clinical attachment level difference at baseline: the difference of mean clinical attachment level between the S group and the NS group at baseline. *P < .05.

4 | DISCUSSION et al., 2001; Stoltenberg et al., 1993). These conflicting results might be caused by different study designs (Grine et al., 2019; Johnson & Periodontitis is a biofilm-induced inflammatory disease caused by Hill, 2004; van Winkelhoff et al., 2001). bacterial dysbiosis and disproportionate host responses (Tonetti It is generally believed that smokers have less favourable clin- et al., 2018). Smoking results in increased susceptibility to periodon- ical outcomes following non-surgical periodontal therapy than titis and less favourable response to periodontal treatment due to non-smokers. Several studies indicated that smokers had less PD the negative effects of chemicals from cigarettes on the interac- reduction or CAL gain than non-smokers (Table 1). The significant tion between host responses and oral microbiota (Lamster, 1992; negative influence of smoking is confirmed by several studies (Ardais Seymour, 1991). et al., 2014; Eltas & Orbak, 2012; Hughes et al., 2006; Jin et al., Smoking has been identified to have long-term negative effects 2000; Palmer et al., 1999; Wan et al., 2009). However, some studies on both local and systemic host immune responses, from innate did not find such significant negative impact of smoking on clinical to adaptive immunity (Mooney et al., 2001; Palmer et al., 2005; outcomes following non-surgical periodontal therapy (Feres et al., Rawlinson et al., 2003; Ryder et al., 1998; Sørensen et al., 2004). 2015; Guru et al., 2018; Preshaw et al., 2013; Türkoğlu et al., 2016). Additionally, nicotine can negatively affect the reparative healing In a study assessing the impact of clinical factors on responses to response by suppressing vascular growth, inhibiting fibroblast pro- non-surgical periodontal therapy, posterior teeth, interproximal liferation and adhesion, as well as collagen production (Haffajee & sites, baseline mobility and deeper initial PD were significantly asso- Socransky, 2001; Tanur et al., 2000; Tipton & Dabbous, 1995). ciated with less PD reduction in periodontitis patients, but smoking Several studies found smoking could affect subgingival microbi- history was not (D’Aiuto et al., 2005). ota (Haffajee & Socransky, 2001; Kanmaz et al., 2019; Mason et al., The current meta-analysis demonstrates that smokers had less 2015; Zambon et al., 1996). The prevalence of periodontal patho- PD reduction and CAL gain following non-surgical periodontal ther- gens, including Porphyromonas gingivalis, Tannerella forsythia and apy compared to non-smokers and that these differences were sta- Aggregatibacter actinomycetemcomitans, in smokers was higher than tistically significant. However, since the mean difference between non-smokers (Huang & Shi, 2019). Additionally, the prevalence of T. groups was 0.33 mm in PD reduction and 0.20 mm in CAL gain, the forsythia was positively associated with the intensity (pack-years) of negative impact of smoking on treatment responses appears modest smoking (Kanmaz et al., 2019; Socransky et al., 1998; Zambon et al., from a clinical perspective. This may explain prior conflicting results 1996). However, some studies did not find significant differences of in the literature. Of note, not all of the selected studies reported periodontal pathogens between smokers and non-smokers (Boström sample size calculations. In studies included in this review, reporting CHANG et al. | 71

FIGURE 3 Subgroup analyses for pooled differences in PD reduction between the smokers with periodontitis (S) and the non-smokers with periodontitis (NS). Subgroup 1: mean baseline probing depth SNS; and subgroup 3: mean baseline probing depth is not available. ES: effect size average PD and CAL changes of all sites may underestimate the true baseline PD was found to have an influence on the final PD reduc- impact of smoking (Eltas & Orbak, 2012; Guru et al., 2018; Palmer tion (Figure 3). et al., 1999; Renvert et al., 1998). When analysing PD reduction in The length of follow-up is a potential factor affecting clinical deep pockets (≥5 mm), the difference in PD reduction between the outcomes. Among all the included studies, nine articles reported two groups was more prominent (0.5 mm). However, there were only follow-up of 1–3 months, seven articles reported follow-up four studies that reported PD reduction in deep pockets (Apatzidou at 6 months, and one article reported follow-up at 12 months. et al., 2005; Preber et al., 1995; Preshaw et al., 2013; Wan et al., Clinically, re-evaluation to assess healing can be done as early 2009). To mitigate the limitations mentioned above, prospective as 4 weeks after non-surgical periodontal therapy (Lowenguth studies specifically designed to evaluate the clinical outcomes of & Greenstein, 1995). Studies reporting results beyond a one- non-surgical periodontal therapy in smokers vs. non-smokers that year follow-up were not included, because most of these lon- are well powered and report both full-mouth and deep site clini- ger-term studies included surgical therapies to treat periodontitis. cal outcomes are needed. The results of the present meta-regres- Additionally, several other confounding factors, such as the oral sion demonstrate that baseline PD was positively associated with hygiene and compliance of patients, rather than smoking it- PD reduction post-treatment. This is not surprising since sites with self, could impact PD and CAL changes (Kwok & Caton, 2007). deep PD generally have more PD reduction following scaling and According to the current analyses, the varied follow-up periods root planning than shallow sites (Cobb, 1996). The impact of smok- up to 1 year were not significantly associated with clinical out- ing on PD reduction remained statistically significant, even though comes of non-surgical periodontal therapy. 72 | CHANG et al.

FIGURE 4 Subgroup analyses for pooled differences in PD reduction between the smokers with periodontitis (S) and the non-smokers with periodontitis (NS). Subgroup 1: follow-up ≤3 months and subgroup 2: follow-up >3 months. ES: effect size

The current systematic review did not evaluate treatment out- validate the smoking exposure (SRNT Subcommittee on Biochemical comes in former smokers. Studies have showed that smoking cessa- Verification, 2002). Nonetheless, only one of the included studies tion can significantly improve clinical outcomes of both non-surgical reported chemical validation of smoking status by measuring the and surgical periodontal treatments (Costa & Cota, 2019; Haber & carbon monoxide exhaled from smokers (Hughes et al., 2006). The Kent, 1992; Preshaw et al., 2005). Smoking intensity and duration impact of smoking duration could not be analysed because only can also affect clinical outcomes of periodontal therapies (Baljoon three of the included studies reported such information (Ardais et al., 2005; Bergström et al., 2000; Kaldahl et al., 1996). Current et al., 2014; Türkoğlu et al., 2016; Wan et al., 2009). Further clinical results did not show that heavy smokers, smoking ≥20 cigarettes studies are needed to evaluate the effect of smoking intensity, du- per day, had worse clinical outcomes (Table 2). However, the results ration and smoking cessation length of time on clinical outcomes of may be biased by the limited number of studies that included heavy periodontal therapy. smokers (Apatzidou et al., 2005; Guru et al., 2018; Jin et al., 2000; Bleeding on probing is a sign of periodontal inflammation (Polson Preber & Bergström, 1986a). Also, the definition of a “heavy smoker” & Caton, 1985) and associated with deep PD (Zimmermann et al., in these articles was different and most of the studies only reported 2015). The current study did not find an association between BOP mean number of cigarettes in smokers. In addition to self-report and WDPD (p = .80) or WDCAL (p = .95). BOP has been shown to be of the number of cigarettes smoked per day, the concentration of altered in smokers, and this may have affected the influence of BOP chemicals can also be measured in various biological specimens to on PD and CAL changes in the current analysis (Grossi et al., 1997). CHANG et al. | 73

The selected studies included patients with aggressive and neversmokers: a randomized controlled clinical trial. Photodiagnosis and Photodynamic Therapy, 25, 247–252. chronic periodontitis, and the severity of periodontitis was not clear Alqaderi, H., Lee, C.-T., Borzangy, S., & Pagonis, T. C. (2016). Coronal in several studies which made evaluating the impact of smoking on pulpotomy for cariously exposed permanent posterior teeth with treatment responses in patients with different severity levels of closed apices: A systematic review and meta-analysis. Journal of periodontitis not possible. The influence of oral hygiene on treat- Dentistry, 44, 1–7. 10.1016/j.jdent.2015.12.005 Apatzidou, D. A., Riggio, M. P., & Kinane, D. F. (2005). Impact of smoking on ment outcomes could not be evaluated since data on plaque levels the clinical, microbiological and immunological parameters of adult were only reported in a few studies (Apatzidou et al., 2005; D’Aiuto patients with periodontitis. Journal of Clinical Periodontology, 32(9), et al., 2005; Eltas & Orbak, 2012; Guru et al., 2018; Jin et al., 2000; 973–983. https://doi.org/10.1111/j.1600-051X.2005.00788.x Preber & Bergström, 1986a; Preber et al., 1995; Türkoğlu et al., Ardais, R., Mario, T. D. G., Jociana, B., Kantorski, K. Z., & Moreira, C. H. 2016). Patients’ compliance can significantly affect the outcomes of C. (2014). The effect of smoking on bleeding on probing after non- surgical periodontal therapy: a quasi-experimental study. Brazilian periodontal therapy, but it was not clearly described in most of the Oral Research, 28, 1-7. 10.1590/1807-3107BOR-2014.vol28.0058 . included studies and, thus, its impact could not be assessed. Finally, Axelsson, P., Paulartder, J., & Lindhe, J. (1998). Relationship between only outcomes within 1 year were analysed; therefore, the present smoking and dental status in 35-, 50-, 65-, and 75-year-old individ- results may not be applicable for longer-term follow-ups. uals. Journal of Clinical Periodontology, 25(4), 297-305. https://doi. org/10.1111/j.1600-051X.1998.tb02444.x Baljoon, M., Natto, S., & Bergström, J. (2005). Long-term effect of smoking on vertical periodontal bone loss. Journal 5 | CONCLUSION of Clinical Periodontology, 32(7), 789–797. https://doi. org/10.1111/j.1600-051X.2005.00765.x Bergström, J., Eliasson, S., & Dock, J. (2000). Exposure to tobacco smok- The present systematic review indicates that smoking has sig- ing and periodontal health. Journal of Clinical Periodontology, 27(1), nificant negative effects on clinical outcomes of non-surgical peri- 61–68. https://doi.org/10.1034/j.1600-051x.2000.027001061.x odontal therapy. Periodontitis in smokers has significantly less PD Boström, L., Bergström, J., Dahlén, G., & Linder, L. E. (2001). reduction and CAL gain than non-smokers within one year after Smoking and subgingival microflora in periodontal disease. Journal of Clinical Periodontology, 28(3), 212–219. treatment. 10.1034/j.1600-051x.2001.028003212.x Boström, L., Linder, L. E., & Bergström, J. (1998). Influence of smok- ACKNOWLEDGEMENTS ing on the outcome of periodontal surgery. Journal of Clinical The authors would like to thank Dr. I. Darby from the University Periodontology, 25(3), 194-201. https://doi.org/10.1111/j.1600- 051X.1998.tb02428.x of Melbourne, Australia; Dr. R. Farina and Dr. L. Trombelli from the Camargo, G. A. D. C. G., Abreu, M. G. L., Cordeiro, R. D. S., Wenderoscky, University of Ferrara, Italy; Dr. J.M. Varghese from the Manipal L. D. F., & Duque, C. (2016). Prevalence of periodontopatho- Academy of Higher Education, India; and Dr. W.E. Ward from gens and Candida spp. in smokers after nonsurgical periodontal the Brock University, Canada, for providing additional informa- therapy – A pilot study. Brazilian Oral Research, 30, https://doi. tion related to their articles. Their time and assistance are greatly org/10.1590/1807-3107BOR-2016.vol30.0092 Chandra, R. V., Sandhya, Y. P., Nagarajan, S., Reddy, B. H., Naveen, A., & appreciated. Murthy, K. R. V. (2012). Efficacy of lycopene as a locally delivered gel in the treatment of chronic periodontitis: smokers vs nonsmok- CONFLICT OF INTEREST ers. Quintessence International, 43(5), 401–411. The authors report no conflict of interest for this study. Chang, C. H., Han, M. L., Teng, N. C., Lee, C. Y., Huang, W. T., Lin, C. T., & Huang, Y. K. (2018). Cigarette smoking aggravates the activity of periodontal disease by disrupting redox homeostasis–An ORCID observational study. Scientific Reports, 8(1), 11055. https://doi. Jennifer Chang https://orcid.org/0000-0001-9315-5500 org/10.1038/s41598-018-29163-6 Chun-Teh Lee https://orcid.org/0000-0001-7812-5637 Christan, C., Dietrich, T., Hägewald, S., Kage, A., & Bernimoulin, J. P. (2002). White blood cell count in generalized aggressive periodontitis after non-surgical therapy. Journal of Clinical Periodontology, 29(3), 201– REFERENCES 206. https://doi.org/10.1034/j.1600-051x.2002.290303.x Abreu, M. G. L., Kawamoto, D., Mayer, M. P. A., Pascoal, V. D. B., Caiaffa, Cobb, C. M. (1996). Non-surgical pocket therapy: Mechanical. Annals K. S., Zuza, E. P., Duque, C., & Camargo, G. A. D. C. G. (2019). of Periodontology, 1(1), 443–490. https://doi.org/10.1902/ Frequency of Porphyromonas gingivalis fimA in smokers and non- annals.1996.1.1.443 smokers after periodontal therapy. Journal of Applied Oral Science, Costa, F. O., & Cota, L. O. M. (2019). Cumulative smoking exposure 27, https://doi.org/10.1590/1678-7757-2018-0205 and cessation associated with the recurrence of periodontitis in Ah, M. K., Johnson, G. K., Kaldahl, W. B., Patil, K. D., & Kalkwart, K. L. periodontal maintenance therapy: A 6-year follow-up. Journal (1994). The effect of smoking on the response to periodontal ther- of Periodontology, 90(8), 856–865. https://doi.org/10.1002/ apy. Journal of Clinical Periodontology, 21(2), 91–97. https://doi. JPER.18-0635 org/10.1111/j.1600-051X.1994.tb00285.x D’Aiuto, F., Ready, D., Parkar, M., & Tonetti, M. S. (2005). Relative con- AlAhmari, F., Ahmed, H. B., Al-Kheraif, A. A., Javed, F., & Akram, Z. tribution of patient-, tooth-, and site-associated variability on the (2019). Effectiveness of scaling and root planning with and with- clinical outcomes of subgingival debridement. I. Probing depths. out adjunct antimicrobial photodynamic therapy in the treat- Journal of Periodontology, 76(3), 398–405. https://doi.org/10.1902/ ment of chronic periodontitis among cigarette-smokers and jop.2005.76.3.398 74 | CHANG et al.

Darby, I. B., Hodge, P. J., Riggio, M. P., & Kinane, D. F. (2005). Clinical of recommendations. BMJ, 336(7650), 924–926. https://doi. and microbiological effect of scaling and root planing in smoker org/10.1136/bmj.39489.470347.AD and non-smoker chronic and aggressive periodontitis patients. Haber, J., & Kent, R. L. (1992). Cigarette smoking in a periodontal practice. Journal of Clinical Periodontology, 32(2), 200–206. https://doi. Journal of Periodontology, 63(2), 100–106. https://doi.org/10.1902/ org/10.1111/j.1600-051X.2005.00644.x jop.1992.63.2.100 Dodington, D. W., Fritz, P. C., Sullivan, P. J., & Ward, W. E. (2015). Higher Haffajee, A. D., & Socransky, S. S. (2001). Relationship of cigarette smok- intakes of fruits and vegetables, β-carotene, vitamin C, α-tocoph- ing to attachment level profiles. Journal of Clinical Periodontology, erol, EPA, and DHA are positively associated with periodontal heal- 28(4), 283–295. https://doi.org/10.1034/j.1600-051x.2001.02800 ing after nonsurgical periodontal therapy in nonsmokers but not in 4283.x smokers. The Journal of nutrition, 145(11), 2512–2519. Harris, R. J. (1994). The connective tissue with partial thickness double Eltas, A., & Orbak, R. (2012). Clinical effects of Nd:YAG laser applications pedicle graft: the results of 100 consecutively-treated defects. during nonsurgical periodontal treatment in smoking and nonsmok- Journal of Periodontology, 65(5), 448–461. https://doi.org/10.1902/ ing patients with chronic periodontitis. Photomedicine and Laser jop.1994.65.5.448 Surgery, 30(7), 360–366. 10.1089/pho.2011.3184 Heasman, L., Stacey, F., Preshaw, P. M., McCracken, G. I., Hepburn, Erdemir, E. O., Duran, I., & Haliloglu, S. (2004). Effects of smok- S., & Heasman, P. A. (2006). The effect of smoking on peri- ing on clinical parameters and the gingival crevicular fluid lev- odontal treatment response: A review of clinical evidence. els of IL-6 and TNF-α in patients with chronic periodontitis. Journal of Clinical Periodontology, 33(4), 241–253. https://doi. Journal of Clinical Periodontology, 31(2), 99–104. https://doi. org/10.1111/j.1600-051X.2006.00902.x org/10.1111/j.0303-6979.2004.00454.x Helal, O., Göstemeyer, G., Krois, J., Fawzy El Sayed, K., Graetz, C., & Farina, R., Simonelli, A., Rizzi, A., & Trombelli, L. (2010). Effect of smoking Schwendicke, F. (2019). Predictors for tooth loss in periodontitis status on pocket probing depth and bleeding on probing following patients: Systematic review and meta-analysis. Journal of Clinical non-surgical periodontal therapy. Minerva Stomatologica, 59(1–2), Periodontology, 46(7), 699–712. https://doi.org/10.1111/jcpe.13118 1–12. http://europepmc.org/abstract/MED/20212405 Hendek, M. K., Erdemir, E. O., Kisa, U., & Ozcan, G. (2015). Effect of Feldman, R. S., Alman, J. E., & Chauncey, H. H. (1987). Periodontal disease initial periodontal therapy on oxidative stress markers in gingival indexes and tobacco smoking in healthy aging men. Gerodontics, 3, cervicular fluid, saliva, and serum in smokers and non-smokers with 43–46. http://ci.nii.ac.jp/naid/10020175708/en/ chronic periodontitis. Journal of Periodontology, 86(2), 273-282. Feres, M., Bernal, M. A. C., Matarazzo, F., Faveri, M., Duarte, P. M., & https://doi.org/10.1902/jop.2014.140338 Figueiredo, L. C. (2015). Subgingival bacterial recolonization after scal- Higgins, J., & Green, S. (2011). Cochrane handbook for systematic re- ing and root planning in smokers with chronic periodontitis. Australian views of interventions. The Cochrane Collaboration. Dental Journal, 60(2), 225–232. https://doi.org/10.1111/adj.12225 Huang, C., & Shi, G. (2019). Smoking and microbiome in oral, airway, gut Follmann, D., Elliott, P., Suh, I., & Cutler, J. (1992). Variance imputa- and some systemic diseases. Journal of Translational Medicine, 17(1), tion for overviews of clinical trials with continuous response. 225. https://doi.org/10.1186/s12967-019-1971-7 Journal of Clinical Epidemiology, 45(7), 769–773. https://doi. Hughes, F. J., Syed, M., Koshy, B., Bostanci, N., McKay, I. J., Curtis, M. org/10.1016/0895-4356(92)90054-Q A., Marcenes, W., & Croucher, R. E. (2006). Prognostic factors in Genco, R. J., & Borgnakke, W. S. (2013). Risk factors for peri- the treatment of generalized aggressive periodontitis: II. Effects of odontal disease. Periodontology 2000, 62(1), 59–94. doi: smoking on initial outcome. Journal of Clinical Periodontology, 33(9), 10.1111/j.1600-0757.2012.00457.x. 671–676. 10.1111/j.1600-051X.2006.00965.x Gleissner, C., Springborn, I., & Willershausen, B. (2003). Effect of peri- Jiao, J., Shi, D., Cao, Z., Meng, H., Lu, R., Zhang, L., Song, Y., & Zhao, J. (2017). odontal therapy on sulcular sulphide level a longitudinal study. Effectiveness of non-surgical periodontal therapy in a large Chinese pop- European Journal of Medical Research, 8(1), 33–46. http://europ ulation with chronic periodontitis. Journal of Clinical Periodontology, 44(1), epmc.org/abstract/MED/12578752 42–50. https://doi.org/10.1111/jcpe.12637 Grine, G., Royer, A., Terrer, E., Diallo, O. O., Drancourt, M., & Aboudharam, Jin, L., Wong, K. Y., Leung, W. K., Leung, W. K., Corbet, E. F., & Corbet, G. (2019). Tobacco smoking affects the salivary gram-positive E. F. (2000). Comparison of treatment response patterns follow- bacterial population. Frontiers in Public Health, 7, 196. https://doi. ing scaling and root planing in smokers and non-smokers with un- org/10.3389/fpubh.2019.00196 treated adult periodontitis. The Journal of Clinical Dentistry, 11(2), Grossi, S. G., Zambon, J., Machtei, E. L. I. E., Schifferle, R., Andreana, 35–41. http://europepmc.org/abstract/MED/11460281 S., Genco, R. J., Cummins, D., & Harrap, G. (1997). Effect of smok- Johnson, G. K., & Hill, M. (2004). Cigarette smoking and the periodon- ing and smoking cessation on healing after mechanical periodontal patient. Journal of Periodontology, 75(2), 196–209. https://doi. tal therapy. The Journal of the American Dental Association, 128(5), org/10.1902/jop.2004.75.2.196 599–607. https://doi.org/10.14219/jada.archive.1997.0259 Kaldahl, W. B., Johnson, G. K., Patil, K. D., & Kalkwarf, K. L. (1996). Levels Gunsolley, J. C., Quinn, S. M., Tew, J., Gooss, C. M., Brooks, C. N., & of cigarette consumption and response to periodontal therapy. Schenkein, H. A. (1998). The effect of smoking on individuals with Journal of Periodontology, 67(7), 675–681. https://doi.org/10.1902/ minimal periodontal destruction. Journal of Periodontology, 69(2), jop.1996.67.7.675 165–170. https://doi.org/10.1902/jop.1998.69.2.165 Kanmaz, B., Lamont, G., Danacı, G., Gogeneni, H., Buduneli, N., & Scott, Guru, S., Sam, S. E., Rajan, S., & Padmanabhan, S. (2018). Comparative D. A. (2019). Microbiological and biochemical findings in relation evaluation of salivary hepatocyte growth factor in smokers and with clinical periodontal status in active smokers, non-smokers non-smokers with chronic periodontitis. Journal of Investigative and and passive smokers. Tobacco Induced Diseases, 17, 1–6. https://doi. Clinical Dentistry, 9(2), e12306. https://doi.org/10.1111/jicd.12306 org/10.18332/tid/104492 Guyatt, G. H., Oxman, A. D., Schünemann, H. J., Tugwell, P., & Knottnerus, Krall, E. A., Garvey, A. J., & Garcia, R. I. (1999). Alveolar bone loss and A. (2011). GRADE guidelines: A new series of articles in the Journal tooth loss in male cigar and pipe smokers. The Journal of the American of Clinical Epidemiology. Journal of Clinical Epidemiology, 64(4), Dental Association, 130(1), 57–64. https://doi.org/10.14219/jada. 380–382. https://doi.org/10.1016/j.jclinepi.2010.09.011 archive.1999.0029 Guyatt, G. H., Oxman, A. D., Vist, G. E., Kunz, R., Falck-Ytter, Y., Kwok, V., & Caton, J. G. (2007). Commentary: Prognosis revisited: A sys- Alonso-Coello, P., & Schünemann, H. J. (2008). GRADE: an tem for assigning periodontal prognosis. Journal of Periodontology, emerging consensus on rating quality of evidence and strength 78(11), 2063-2071. https://doi.org/10.1902/jop.2007.070210 CHANG et al. | 75

Lamster, I. B. (1992). The host response in gingival crevicular fluid: fluid levels of transforming growth factor-beta 1, tissue plasmin- Potential applications in periodontitis clinical trials. Journal of ogen activator and plasminogen activator inhibitor 1 in smoking Periodontology, 63(12S), 1117–1123. https://doi.org/10.1902/ and non-smoking. Journal of Periodontal Research, 52(5), 872–882. jop.1992.63.12s.1117 https://doi.org/10.1111/jre.12457 Lee, C.-T., Hum, L., & Chen, Y.-W. (2016). The effect of regenerative Papapanou, P. N., Sanz, M., Buduneli, N., Dietrich, T., Feres, M., Fine, periodontal therapy in preventing periodontal defects after the D. H., Flemmig, T. F., Garcia, R., Giannobile, W. V., Graziani, F., extraction of third molars: A systematic review and meta-analysis. Greenwell, H., Herrera, D., Kao, R. T., Kebschull, M., Kinane, D. F., The Journal of the American Dental Association, 147(9), 709–719.e4. Kirkwood, K. L., Kocher, T., Kornman, K. S., Kumar, P. S., … Tonetti, https://doi.org/10.1016/j.adaj.2016.03.005 M. S. (2018). Periodontitis: Consensus report of workgroup 2 of Liberati, A., Altman, D. G., Tetzlaff, J., Mulrow, C., Gøtzsche, P. C., the 2017 world workshop on the classification of periodontal and Ioannidis, J. P. A., Clarke, M., Devereaux, P. J., Kleijnen, J., & Moher, peri-implant diseases and conditions. Journal of Periodontology, D. (2009). The PRISMA statement for reporting systematic reviews 89(S1), S173-S182. https://doi.org/10.1002/JPER.17-0721 and meta-analyses of studies that evaluate health care interven- Phillips, B., Ball, C., Badenoch, D., Straus, S., Haynes, B., & Dawes, M. tions: explanation and elaboration. PLoS Med, 6(7), e1000100. (2011). Oxford centre for evidence-based medicine levels of evi- https://doi.org/10.1371/journal.pmed.1000100 dence. BJU International, 107(5), 870. Lowenguth, R. A., & Greenstein, G. (1995). Clinical and microbiolog- Polson, A. M., & Caton, J. G. (1985). Current status of bleeding in the ical response to nonsurgical mechanical periodontal therapy. diagnosis of periodontal diseases. Journal of Periodontology, 56, 1–3. Periodontology 2000, 9(1), 14–22. doi: 10.1111/j.1600-0757.1995. 10.1902/jop.1985.56.11s.1 tb00052.x. Preber, H., & Bergström, J. (1986a). The effect of non-surgical treatment Machtei, E. E., Hausmann, E., Schmidt, M., Grossi, S. G., Dunford, R., on periodontal pockets in smokers and non-smokers. Journal of Schifferle, R., Munoz, K., Davies, G., Chandler, J., & Genco, R. J. Clinical Periodontology, 13(4), 319–323. https://doi.org/10.1111/ (1998). Radiographic and clinical responses to periodontal therapy. j.1600-051X.1986.tb02229. Journal of Periodontology, 69(5), 590–595. https://doi.org/10.1902/ Preber, H., & Bergström, J. (1986b). Effect of non-surgical treatment on jop.1998.69.5.590 gingival bleeding in smokers and non-smokers. Acta Odontologica Mason, M. R., Preshaw, P. M., Nagaraja, H. N., Dabdoub, S. M., Rahman, Scandinavica, 44(2), 85–89. https://doi.org/10.3109/0001635860 A., & Kumar, P. S. (2015). The subgingival microbiome of clinically 9041312 healthy current and never smokers. The ISME Journal, 9(1), 268– Preber, H., & Bergström, J. (1990). Effect of cigarette smoking on peri- 272. https://doi.org/10.1038/ismej.2014.114 odontal healing following surgical therapy. Journal of Clinical Meulman, T., Casarin, R. C. V., Peruzzo, D. C., Giorgetti, A. P., Barbagallo, Periodontology, 17(5), 324–328. https://doi.org/10.1111/j.1600- A., Casati, M. Z., Sallum, E. A., Gonçalves, R. B., & Jr, F. H. N. (2012). 051X.1990.tb01098.x Impact of supragingival therapy on subgingival microbial profile in Preber, H., Linder, L., & Bergström, J. (1995). Periodontal healing and smokers versus non-smokers with severe chronic periodontitis. periopathogenic microflora in smokers and non-smokers. Journal of Journal of Oral Microbiology, 4(1), 8640. https://doi.org/10.3402/ Clinical Periodontology, 22(12), 946-952. https://doi.org/10.1111/ jom.v4i0.8640 j.1600-051X.1995.tb01800.x Mohan, R., Varghese, J., Bhat, V., & Chianeh, Y. R. (2019). The effect of Preshaw, P. M., Heasman, L., Stacey, F., Steen, N., McCracken, G. I., & nonsurgical periodontal therapy on pentraxin 3 levels in smokers Heasman, P. A. (2005). The effect of quitting smoking on chronic and nonsmokers with chronic periodontitis. General Dentistry, 67(2), periodontitis. Journal of Clinical Periodontology, 32(8), 869–879. e1–e6. http://europepmc.org/abstract/MED/30875312 https://doi.org/10.1111/j.1600-051X.2005.00779.x Moher, D., Shamseer, L., Clarke, M., Ghersi, D., Liberati, A., Petticrew, Preshaw, P. M., Holliday, R., Law, H., & Heasman, P. A. (2013). Outcomes M., Shekelle, P., & Stewart, L. A. & Group, P.-P (2015). Preferred of non-surgical periodontal treatment by dental hygienists in train- reporting items for systematic review and meta-analysis protocols ing: impact of site- and patient-level factors. International Journal of (PRISMA-P) 2015 statement. Systematic Reviews, 4(1), 1. https:// Dental Hygiene, 11(4), 273–279. https://doi.org/10.1111/idh.12032 doi.org/10.1186/2046-4053-4-1 Preshaw, P. M., Lauffart, B., Zak, E., Jeffcoat, M. K., Barton, I., & Montero, E., Herrera, D., Sanz, M., Dhir, S., Van Dyke, T., & Sima, C. (2019). Heasman, P. A. (1999). Progression and treatment of chronic adult Development and validation of a predictive model for periodontitis periodontitis. Journal of Periodontology, 70(10), 1209–1220. https:// using NHANES 2011–2012 data. Journal of Clinical Periodontology, doi.org/10.1902/jop.1999.70.10.1209 46(4), 420–429. https://doi.org/10.1111/jcpe.13098 Preus, H. R., Sandvik, L., Gjermo, P., & Baelum, V. (2014). Baseline ad- Mooney, J., Hodge, P. J., & Kinane, D. F. (2001). Humoral immune re- justment and change revisited: effect of smoking on change in peri- sponse in early-onset periodontitis: Influence of smoking. Journal odontal status following periodontal therapy. European Journal of of Periodontal Research, 36(4), 227–232. https://doi.org/10.103 Oral Sciences, 122(2), 89–99. https://doi.org/10.1111/eos.12111 4/j.1600-0765.2001.036004227.x Pucher, J. J., Shibley, O., Dentino, A. R., & Ciancio, S. G. (1997). Results Nassrawin, N. A. (2010). Effect of smoking on the response to nonsurgi- of limited initial periodontal therapy in smokers and non-smokers. cal periodontal therapy. Eastern Mediterranean Health Journal, 16(2), Journal of Periodontology, 68(9), 851–856. https://doi.org/10.1902/ 162–165. jop.1997.68.9.851 Palmer, R. M., Matthews, J. P., & Wilson, R. F. (1999). Non-surgical peri- Rawlinson, A., Grummitt, J. M., Walsh, T. F., & Ian Douglas, C. W. odontal treatment with and without adjunctive metronidazole in (2003). Interleukin 1 and receptor antagonist levels in gin- smokers and non-smokers. Journal of Clinical Periodontology, 26(3), gival cervicular fluid in heavy smokers versus non-smokers. 158-163. https://doi.org/10.1034/j.1600-051X.1999.260305.x Journal of Clinical Periodontology, 30(1), 42–48. https://doi. Palmer, R. M., Wilson, R. F., Hasan, A. S., & Scott, D. A. (2005). org/10.1034/j.1600-051X.2003.300107.x Mechanisms of action of environmental factors – tobacco smok- Renvert, S., Dahlén, G., & Wikström, M. (1998). The clinical and micro- ing. Journal of Clinical Periodontology, 32(s6), 180-195. https://doi. biological effects of non-surgical periodontal therapy in smokers org/10.1111/j.1600-051X.2005.00786.x and non-smokers. Journal of Clinical Periodontology, 25(2), 153–157. Pamuk, F., Lütfioğlu, M., Aydoğdu, A., Koyuncuoglu, C. Z., Cifcibasi, E., & https://doi.org/10.1111/j.1600-051X.1998.tb02421.x Badur, O. S. (2017). The effect of low-level laser therapy as an ad- Ryder, M. I., Fujitaki, R., Lebus, S., Mahboub, M., Faia, B., junct to non-surgical periodontal treatment on gingival cervicular Muhaimin, D., Hamada, M., & Hyun, W. (1998). Alterations of 76 | CHANG et al.

neutrophil L-selection and CD18 expression by tobacco smoke: im- Van Winkelhoff, A. J., Bosch-Tijhof, C. J., Winkel, E. G., & van der Reijden, plications for periodontal diseases. Journal of Periodontal Research, W. A. (2001). Smoking affects the subgingival microflora in peri- 33(6), 359–368. https://doi.org/10.1111/j.1600-0765.1998.tb022 odontitis. Journal of Periodontology, 72(5), 666–671. https://doi. 11.x org/10.1902/jop.2001.72.5.666 Seymour, G. J. (1991). Importance of the host response in the periodon- Wan, C. P., Leung, W. K., Wong, M. C. M., Wong, R. M. S., Wan, P., Lo, tium. Journal of Clinical Periodontology, 18(6), 421–426. https://doi. E. C. M., & Corbet, E. F. (2009). Effects of smoking on healing re- org/10.1111/j.1600-051X.1991.tb02310.x sponse to non-surgical periodontal therapy: a multilevel modelling Socransky, S. S., Haffajee, A. D., Cugini, M. A., Smith, C., & Kent, R. L. Jr analysis. Journal of Clinical Periodontology, 36(3), 229–239. https:// (1998). Microbial complexes in subgingival plaque. Journal of Clinical doi.org/10.1111/j.1600-051X.2008.01371.x Periodontology, 25(2), 134–144. https://doi.org/10.1111/j.1600- Wells, G. A., Shea, B., O’Connell, D., Peterson, J., Welch, V., Losos, M., & 051X.1998.tb02419.x Tugwell, P. (2000). The Newcastle-Ottawa Scale (NOS) for assess- Sørensen, L. T., Nielsen, H. B., Kharazmi, A., & Gottrup, F. (2004). Effect of ing the quality of nonrandomised studies in meta-analyses. http:// smoking and abstention on oxidative burst and reactivity of neu- www.ohri.ca/programs/clinical_epidemiology/oxford.asp trophils and monocytes. Surgery, 136(5), 1047–1053. 10.1016/j. Weyant, R. J. (1994). Characteristics associated with the loss and peri-im- surg.2004.04.001 plant tissue health of endosseous dental implants. The International SRNT Subcommittee on Biochemical Verification. (2002). Biochemical ver- Journal of Oral & Maxillofacial Implants, 9, 95–102. ification of tobacco use and cessation. Nicotine & Tobacco Research, Zambon, J. J., Grossi, S. G., Machtei, E. E., Ho, A. W., Dunford, R., 4(2), 149–159. https://doi.org/10.1080/14622200210123581 & Genco, R. J. (1996). Cigarette smoking increases the risk for Stoltenberg, J. L., Osborn, J. B., Pihlstrom, B. L., Herzberg, M. C., Aeppli, subgingival infection with periodontal pathogens. Journal of D. M., Wolff, L. F., & Fischer, G. E. (1993). Association between ciga- Periodontology, 67(10S), 1050–1054. https://doi.org/10.1902/ rette smoking, bacterial pathogens, and periodontal status. Journal jop.1996.67.10s.1050 of Periodontology, 64(12), 1225–1230. https://doi.org/10.1902/ Zimmermann, H., Hagenfeld, D., Diercke, K., El-Sayed, N., Fricke, J., jop.1993.64.12.1225 Greiser, K. H., Kühnisch, J., Linseisen, J., Meisinger, C., Pischon, Tanur, E., McQuade, M. J., McPherson, J. C., Al-Hashimi, I. H., & Rivera- N., Pischon, T., Samietz, S., Schmitter, M., Steinbrecher, A., Kim, Hidalgo, F. (2000). Effects of nicotine on the strength of attach- T.-S., & Becher, H. (2015). Pocket depth and bleeding on probing ment of gingival fibroblasts to glass and non-diseased human root and their associations with dental, lifestyle, socioeconomic and surfaces. Journal of Periodontology, 71(5), 717–722. https://doi. blood variables: a cross-sectional, multicenter feasibility study of org/10.1902/jop.2000.71.5.717 the German National Cohort. BMC Oral Health, 15(1), 7. https://doi. Tipton, D. A., & Dabbous, M. K. (1995). Effects of nicotine on prolifera- org/10.1186/1472-6831-15-7 tion and extracellular matrix production of human gingival fibro- Zuabi, O., Machtei, E. E., Ben-Aryeh, H., Ardekian, L., Peled, M., & Laufer, blasts in vitro. Journal of Periodontology, 66(12), 1056–1064. https:// D. (1999). The effect of smoking and periodontal treatment on sali- doi.org/10.1902/jop.1995.66.12.1056 vary composition in patients with established periodontitis. Journal Tonetti, M. S., Greenwell, H., & Kornman, K. S. (2018). Staging and grad- of Periodontology, 70(10), 1240–1246. https://doi.org/10.1902/ ing of periodontitis: Framework and proposal of a new classifica- jop.1999.70.10.1240 tion and case definition. Journal of Clinical Periodontology, 45(S20), S149–S161. https://doi.org/10.1111/jcpe.12945 Tonetti, M. S., Pini-Prato, G., & Cortellini, P. (1995). Effect of cigarette SUPPORTING INFORMATION smoking on periodontal healing following GTR in infrabony de- Additional supporting information may be found online in the fects. Journal of Clinical Periodontology, 22(3), 229–234. https://doi. Supporting Information section. org/10.1111/j.1600-051X.1995.tb00139.x Türkoğlu, O., Eren, G., Emingil, G., Azarsız, E., Kutukculer, N., & Atilla, G. (2016). Does smoking affect gingival crevicular fluid LL-37 levels following non-surgical periodontal treatment in chronic periodon- How to cite this article: Chang J, Meng HW, Lalla E, Lee CT. titis? Archives of Oral Biology, 61, 98–105. https://doi.org/10.1016/j. The impact of smoking on non-surgical periodontal therapy: archoralbio.2015.10.018 A systematic review and meta-analysis. J Clin Periodontol. Van Der Weijden, G. A., De Slegte, C., Timmerman, M. F., & Van Der Velden, U. (2001). Periodontitis in smokers and non-smokers: intra-oral dis- 2021;48:61–76. https://doi.org/10.1111/jcpe.13384 tribution of pockets. Journal of Clinical Periodontology, 28(10), 955– 960. https://doi.org/10.1034/j.1600-051x.2001.028010955.x Received: 5 March 2020 | Revised: 9 August 2020 | Accepted: 6 October 2020 DOI: 10.1111/jcpe.13389

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

Peripheral T helper cell profiles during management of periodontitis

1Melbourne Dental School, The University of Melbourne, Carlton, VIC, Australia Abstract 2Centre for Oral Health Research, Aim: Periodontitis has been associated with other systemic diseases with underlying Melbourne Dental School, The University of Melbourne, Carlton, VIC, Australia inflammation responsible for the shared link. This study evaluated longitudinal varia- 3Austin Hospital, Heidelberg, VIC, tion in peripheral T helper cells in periodontitis patients undergoing management over Australia 1 year.

Correspondence Materials and methods: Periodontal parameters and peripheral blood mononuclear Ivan B. Darby, Melbourne Dental School, cells (PBMCs) were collected from 54 periodontitis patients at baseline, and 3-, 6- and The University of Melbourne, 720 + + + + Swanston Street, Carlton, VIC 3053, 12-months post-treatment and 40 healthy controls. IFN-γ , IL-4 , IL-17 and Foxp3 Australia. + + and their double-positive expression were identified in CD4 and TCRαβ cells using Email: [email protected] flow cytometry. PBMCs were incubated with P. gingivalis, and IFN-γ, IL-4, IL-17 and Funding information This study was supported by grants from IL-10 in cell supernatant were measured by ELISA. Cells and cytokines were also as- the National Health and Medical Research sessed based on clinical response to treatment where good (<10% of sites), moderate Council (APP1029878) and the Australian Dental Research Foundation (ADRF, 211- (10–20%) and poor (>20%) treatment outcome (TxO) groups had probing depths of 2017). ≥5 mm at study conclusion. + Results: IFN-γ cells were lower at baseline, and 3- and 6-months compared to health, whereas Foxp3+ cells were increased at 12-months compared to all preceding time- + points and health. The good TxO group showed treatment-related variation in IFN-γ + and Foxp3 cells, whereas the poor TxO group did not. IFN-γ and IL-17 cytokine expression in cell supernatants was significantly lower at baseline compared to health, and IFN-γ and IL-10 showed treatment-related decrease. + + Conclusion: This study suggests that IFN-γ and Foxp3 cells may have a role in the systemic compartment in periodontitis. Periodontal management has local and systemic effects, and thus, assessment and management of periodontitis should form an integral part of overall systemic health.

KEYWORDS cytokines, periodontal debridement, periodontitis, T lymphocytes, helper-induced, T lymphocytes, regulatory

wileyonlinelibrary.com/journal/jcpe | 77 J Clin Periodontol. 2021;48:77–91. 78 | MEDARA et al.

1 | INTRODUCTION Clinical Relevance Periodontitis is a polymicrobial immune-inflammatory disease af- Scientific rationale for study: Periodontitis is linked to other fecting the supporting structures of a tooth. Although localized to chronic diseases such as diabetes and cardiovascular dis- the oral cavity, mounting evidence points to a role for periodontitis eases. This link is thought to be due to the leaking of bacte- as a risk factor for several systemic diseases including atheroscle- ria and inflammatory products which affects areas distant rotic cardiovascular diseases (Dietrich et al., 2013) and type 2 diabe- to the mouth. Ultimately, the host immune response de- tes (Borgnakke et al., 2013). This association is thought to be related termines disease aetiology and clinical outcome in peri- to systemic dissemination of periodontal bacteria or inflammatory odontitis. T helper (Th) cells are known to play a role in mediators and co-localization of risk factors such as increasing age, disease pathogenesis. It remains unclear as to what effect stress and smoking behaviours (Bui et al., 2019; Konkel et al., 2019), treatment has on peripheral Th cells during disease man- and is corroborated by animal models (Blasco-Baque et al., 2017; agement longitudinally. Jain et al., 2003). In periodontitis, the chronic non-resolving inflam- Principal findings: IFN- + cells were lower in periodontitis mation and dysbiotic microbiota lead to ulceration of the highly γ compared to health. Foxp3+ cells were higher at 12-months vascularized oral epithelial barrier with greater access to systemic post-treatment. The good treatment outcome group circulation which may lead to adverse effects at distal tissue sites showed treatment-related changes in IFN- +- and Foxp3+- or subversion and/or amplification of the systemic inflammatory re- γ expressing cells, whereas the poor treatment outcome sponse (Konkel et al., 2019). group did not. The mechanisms by which the host develops periodontitis remain Practical implications: Our results provide a broad under- unclear. However, disease pathogenesis is at least, in part, mediated standing of T cells in peripheral circulation and establish by the actions of T cells and cytokines. Upon antigenic stimulation, for the first time the relative roles of IFN- +, IL-4+, IL-17+ naïve CD4 T helper (Th) cells activate, differentiate and proliferate γ and Foxp3+ cells up to 12-months during management of into distinct effector cells characterized by the production of a sig- periodontitis. Treatment has an effect on some Tcell sub- nature cytokine where Th1, Th2, Th17 and T regulatory (Treg) cells sets in peripheral blood, and management of periodontitis produce interferon (IFN)-γ, interleukin (IL)-4, IL-17 and IL-10, re- may have beneficial effects not only in the oral cavity but spectively. Both protective and pathogenic roles for these Th cells also for overall systemic health. and their cytokines are described in periodontitis (Hajishengallis & Korostoff, 2017). This functional heterogeneity serves to target cells to necessary sites and define the class of immune and tissue responses appropriate for a particular pathogen. reduction (Zhao et al., 2011). Therefore, this study aimed to inves- + + + + Porphyromonas gingivalis may be described as a “keystone patho- tigate the longitudinal variation of IFN-γ , IL-4 , IL-17 and Foxp3 + + gen” in periodontitis, where it is thought to be an essential compo- cells and their double-positive expression in CD4 and TCRαβ cells nent of plaque biofilm altering the local immune microenvironment in PBMCs during management of periodontitis up to 12-months allowing subsequent colonization and overgrowth of pathobionts post-treatment. Furthermore, IFN-γ, IL-4, IL-17 and IL-10 production leading to disease (Hajishengallis et al., 2011). P. gingivalis interac- from PBMCs in response to P. gingivalis was also evaluated. Finally, tion with immune cells may direct the response towards a particular differences based on response to treatment were evaluated. A Th cell lineage and effector function. The presence of IFN-γ in the greater understanding of Tcell profiles will improve our knowledge microenvironment skews towards a Th1 phenotype and inhibits Th2 of the aetio-pathogenesis of periodontitis and lead to targeted treat- (Mosmann et al., 1986) and Th17 (Harrington et al., 2005) cell differ- ment modalities. entiation. Likewise, the presence of other signature cytokines may induce polarization towards a particular lineage and regulate differentiation into others (Cooney et al., 2011; Grünig et al., 1997; Toh 2 | METHODS et al., 2009). Altogether, there is a consensus that local microbial dysbiosis 2.1 | Study population and ensuing immune response in periodontitis may have distant systemic effects (Chapple & Genco, 2013; Tonetti & Van Dyke, 2013). The Human Ethics Sub-Committee, The University of Melbourne Previous literature investigating Th cells in peripheral blood mono- (1339812.3), and the Human Ethics Research Committee, Dental nuclear cells (PBMCs) is conflicting with studies reporting higher Th1 Health Services Victoria (279), approved the study. Sample size (Chen et al., 2016; Schmidt et al., 2014), Th2 (Aoyagi et al., 1995; calculations were based on previously published work by the same Lima et al., 2011), Th17 (Chen et al., 2016; Luo et al., 2014; Schmidt group (Byrne et al., 2009). Informed written consent was gained et al., 2014) and Treg (Sabarish et al., 2016) cells in periodontitis or from each subject at the commencement of the study. Data collec- no significant differences (Cheng et al., 2018; Gemmell & Seymour, tion was executed between 2015 and 2017. The study design and 1998; Lalla et al., 2007; Okada et al., 2017) or treatment-related recruitment process are displayed in Appendix S1, Figure A1. MEDARA et al. | 79

Periodontitis subjects (n = 54) were recruited from the specialist Fixable viability stain BV510 (1 µl) was added and the cells were in- periodontics clinic at The Royal Dental Hospital of Melbourne and cubated at 37°C for 5 min protected from light, after which they Melbourne Dental Clinic, The University of Melbourne. Inclusion were washed once (2 ml azide-free FACS wash (2% w/v BSA, 2 mm criteria for the periodontitis group were at least two non-adjacent EDTA in PBS), 800RCF, 5 min, 20°C). PBMCs were resuspended in sites per quadrant exhibiting probing depths (PD) ≥5 mm, excluding 100 µl azide-free FACS wash, and intra-cellular cytokine staining was the third molars (Darby et al., 2001). Subjects were >21 years, had a performed using eBioscience™ Foxp3/Transcription Factor Staining minimum of 16 teeth (excluding third molars) and were systemically Buffer Set (Thermo Fisher Scientific) following the manufacturer's healthy with no periodontal treatment or antibiotic use within the instructions with pre-diluted antibodies against CD25-BV650 (4 µl, preceding 6 months. Exclusion criteria included pregnancy or lacta- clone-M-A251), TCRαβ-BV785 (4 µl, clone-T10B9.1A-31), CD4- tion and medical conditions affecting the progression of periodon- APCCy7 (4 µl, clone-RPA-T4), IL-17-AF488 (15 µl, clone-N49-653), titis (e.g. diabetes) or requiring pre-medication prior to treatment. IFN-γ-PerCPCy5.5 (4 µl, clone-B27), Foxp3-AF647 (15 µl, clone- For the control group (n = 40), the inclusion criteria were a gen- 259D/C7) and IL-4-PE (20 µl, clone-8D4-8). Samples were ac- der and age (±5 years) match for the periodontitis subject with no quired with BD LSRFortessaTM X-20 (BD Bioscience) and analysed PD >4 mm or percentage of sites with bleeding on probing (%BOP) using FlowJoTM Software (v10, Tree Star Inc.). All antibodies were >30%. The remaining inclusion and exclusion criteria were the same purchased from BD Bioscience, and volumes used were optimized as for the periodontitis group. based on manufacture's recommendations. The gating strategy is The periodontal parameters, mean PD and %BOP, and blood displayed in Figure 1. were collected from periodontitis subjects at baseline before treatment, and at 3-, 6- and 12-months post-treatment, and once from healthy controls. PD and BOP were recorded at six sites around each 2.3 | Cytokine detection tooth using a Williams probe (Hu-Friedy Mfg. Co.). PD was measured to the nearest millimetre from the base of the gingival sulcus to the Fresh PBMCs (1 × 106 cells) were either incubated alone, with 5 µg/ free gingival margin. BOP was assessed visually following probing to ml P. gingivalis W50 whole-cell lysate or 5 µg/ml concanavalin A the base of the pocket up to 30 s after probing. (ConA) (Sigma-Aldrich) for 48 h at 37°C, 5% CO2 in 2 ml cRPMI. Periodontal treatment involved patient education about peri- P. gingivalis was obtained from the Oral Health CRC, Melbourne odontitis and oral hygiene instruction (OHI) prior to non-surgi- Dental School, The University of Melbourne, and was grown and cal quadrant scaling and root debridement usually over a period processed as previously described (Lam et al., 2015). The aspirated of 4 weeks. They were re-evaluated at 3, 6, 9 and 12 months. At supernatant was centrifuged at 800RCF, 5 min, 20°C, aliquoted and re-evaluation, clinical parameter was charted, the outcome of treat- stored at −80°C until analysis. Samples were analysed in duplicate ment evaluated, patients given OHI as needed, and residual pockets and pre-matched ELISA kits were used for IFN-γ (BD Bioscience), re-debrided or surgery performed (n = 5) as necessary. The treat- IL-4 (BD Bioscience), IL-17 (Thermo Fisher Scientific) and IL-10 (BD ment goal was PD <4 mm and %BOP <30%. Bioscience) quantification following the manufacturer's instructions. At the end of one year or when periodontitis patients exited the study, they were assigned to a treatment outcome (TxO) group where the good TxO group had <10%, moderate TxO group had be- 2.4 | Statistical analysis tween 10 and 20%, and poor TxO group had >20% of sites with PD ≥5 mm. Periodontitis subjects (n = 8) who only contributed to base- Independent samples t tests with unequal variances were used for line were not assigned a TxO group. comparison between health and periodontitis at each timepoint. Linear mixed models (LMMs) were used to determine changes with treatment in the mean of the outcome variable over timepoint and 2.2 | Flow cytometry for Tcell phenotyping TxO group in periodontitis. The base model for variation over timepoint included timepoint, age and gender as fixed effects and sub- Peripheral venous blood was collected in potassium-EDTA tubes ject as random effect. The base model for variations over TxO group (BD Biosciences), and PBMCs were isolated by density-gradient included TxO group, timepoint, interaction between TxO group and centrifugation using Histopaque-1077 (Sigma-Aldrich) following timepoint, age and gender as fixed effects and subject as random manufacturer's instructions. PBMCs were used straightaway or cry- effect. To assess the effect of periodontal management, mean PD opreserved for later analysis. and %BOP were added as fixed effects to the base model. Pairwise PBMCs (1 × 106 cells) were diluted in 1 ml cRPMI (RPMI-1640, comparisons of estimated marginal means were used to assess sig- 10% v/v heat-inactivated FBS, 100 U/ml Penicillin-Streptomycin, nificant differences. Visual inspection of residual plots did not show 2 mm L-glutamine) and stimulated with PMA (5 ng/ml) and ionomy- any obvious deviations from normality. cin (2 µm) for 6 h at 37°C, 5% CO2 with brefeldin A (20 µg/ml) added Spearman's correlations were used for associations with after 2 h. After stimulation, PBMCs were washed twice (8 ml PBS, the periodontal parameters at each timepoint. Paired samples 800RCF, 5 min, 20°C) and resuspended to 1 × 106 cells/ml in PBS. t tests were used for comparison between cell phenotypes and 80 | MEDARA et al.

+ + + + FIGURE 1 Gating strategy for phenotyping IFN-γ , IL-4 , IL-17 and Foxp3 expressing T cells. Lymphocytes were identified based on forward scatter area (FSC-A) and side scatter area (SSC-A). Live cells were identified based on the exclusion of viability stain. The lymphocyte population was better identified based on FSC-A and SSC-A, and doublets were excluded using FSC-A and forward scatter + + + + height (FSC-H) followed by side scatter height (SSC-H) and SSC-A. IFN-γ , IL-4 , IL-17 and Foxp3 cells in the total population were + − + + + − − − + + identified and divided based on CD4 TCRαβ , CD4 TCRαβ , TCRαβ CD4 or CD4 TCRαβ expression. CD4 TCRαβ cells were further divided + + + + + + + + + + + + + + into IFNγ IL-4 , IFNγ Foxp3 , IL-4 Foxp3 , IL-4 IL-17 , IFNγ IL-17 , Foxp3 IL-17 and CD25 Foxp3 cells MEDARA et al. | 81 cytokine production with the various stimulation methods within mean PD was significantly different at baseline and 12-months be- each timepoint. tween all three TxO groups, with the poor group showing higher The level of significance was set at p ≤ 0.05 for all variables. mean PD followed by moderate and good TxO groups. At 3- and All statistical analyses were performed in SPSS (v23, IBM SPSS 6-months, mean PD in the poor TxO group was significantly higher Statistics for Windows), and graphs were prepared using GraphPad compared to good and moderate groups. There were no differences Prism (v5, GraphPad Software). in %BOP between the TxO groups at baseline. However, %BOP was significantly higher in the poor compared to the good TxO group at all timepoints post-treatment. 3 | RESULTS

3.1 | Study cohort 3.2 | IFN-γ, IL-4, IL-17 and Foxp3 expression in PBMCs The cohort demographics and periodontal parameters are displayed + in Table 1. Mean PD and %BOP were significantly higher at baseline In both healthy and periodontitis groups, IFN-γ cells were the domi- compared to health and decreased at subsequent timepoints post- nant cell type whereas IL-17+ cells were the minority. When sepa- + + + treatment compared to baseline. When separated into TxO groups, rated based on TCRαβ and CD4 expression, the majority of IFN-γ

TABLE 1 Cohort demographics and periodontal parameters.

Periodontitis

Health (n = 40) Baseline 3-months 6-months 12-months

Gender (n, % male) 14 (35%) 20 (37%) (n = 54) Age at baseline (years, 49.30 ± 10.62 53.28 ± 11.44 mean ±SD) (n = 54) Periodontal parameters Mean PD (mm, mean 2.80 ± 0.28 3.72 ± 0.74a 3.01 ± 0.51a 2.92 ± 0.55a 2.80 ± 0.44a ±SD) (n = 54) (n = 46) (n = 44) (n = 37) %BOP (%, mean ±SD) 10.86 ± 7.72 51.61 ± 25.57a 28.35 ± 15.17a 25.82 ± 18.84a 24.66 ± 18.05a (n = 54) (n = 46) (n = 44) (n = 37) Periodontal parameters separated based on TxO groups Mean PD (mm, mean ±SD) Good TxO group 2.80 ± 0.28 3.22 ± 0.38a 2.66 ± 0.29a 2.49 ± 0.23a 2.41 ± 0.20a (n = 18) (n = 18) (n = 18) (n = 14) Moderate TxO group 2.80 ± 0.28 3.84 ± 0.76a 2.99 ± 0.30a 2.96 ± 0.33a 2.83 ± 0.21a (n = 13) (n = 13) (n = 12) (n = 12) Poor TxO group 2.80 ± 0.28 4.22 ± 0.77a 3.53 ± 0.50a 3.48 ± 0.45a 3.26 ± 0.36a (n = 15) (n = 15) (n = 14) (n = 11) %BOP (%, mean ±SD) Good TxO group 10.86 ± 7.72 47.95 ± 22.94a 20.29 ± 11.85a 17.86 ± 11.37a 16.67 ± 16.55a (n = 18) (n = 18) (n = 18) (n = 14) Moderate TxO group 10.86 ± 7.72 53.17 ± 31.22a 30.95 ± 15.79a 27.26 ± 20.40b 26.62 ± 15.87b (n = 13) (n = 13) (n = 12) (n = 12) Poor TxO group 10.86 ± 7.72 53.62 ± 24.61a 38.92 ± 12.92a 38.27 ± 22.67a 36.11 ± 18.18a (n = 15) (n = 15) (n = 14) (n = 11)

Note: Periodontitis subjects (n = 8) who only contributed to baseline were not assigned to a TxO group. Abbreviations: %BOP, percentage of sites exhibiting bleeding on probing; mm, millimetres; PD, probing depth; SD, standard deviation; TxO, treatment outcome. aSignificant difference to health. bSignificant difference to baseline. cSignificant difference to 3-months post-treatment. dSignificant difference between good and moderate treatment outcome groups at same timepoint. eSignificant difference between good and poor treatment outcome groups at same timepoint. fSignificant difference between moderate and poor treatment outcome groups at same timepoint. 82 | MEDARA et al.

+ + + + FIGURE 2 IFN-γ , IL-4 , IL-17 and Foxp3 cells in PBMCs and their CD4 and TCRαβ expression. Clustered bar charts showing mean and + + + + + − + + + − − − SD for IFN-γ , IL-4 , IL-17 and Foxp3 cells as a percentage of PBMCs (a) and CD4 TCRαβ , TCRαβ CD4 , TCRαβ CD4 and TCRαβ CD4 + + + + expression in IFN-γ , IL-4 , IL-17 and Foxp3 cells (b) in health and periodontitis at baseline, and 3-, 6- and 12-months. *above the bars represents significant differences (p ≤ 0.05) to other cell types within each timepoint. †next to the title represents significant differences (p ≤ 0.05) to other cell types within each group at all timepoints. Analysis of significance was calculated using paired samples t test within each timepoint.

+ + − + + and IL-4 cells were TCRαβ CD4 followed by TCRαβ CD4 cells, baseline compared to health; however, no changes were observed + + + + whereas the majority of IL-17 and Foxp3 cells were TCRαβ CD4 post-treatment. cells (Figure 2). Total Foxp3+ cells (Figure 3b) were significantly higher at + Total IFN-γ cells (Figure 3a) were significantly lower at base- 12-months post-treatment compared to health and all preced- line, and 3- and 6-months in periodontitis compared to health. ing timepoints in periodontitis. When separated into TxO groups, When separated into TxO groups, within the good TxO group, Foxp3+ cells were significantly increased at 12-months compared to + IFN-γ cells were significantly lower at 3-, 6- and 12-months all timepoints in periodontitis within the good TxO group and signifi- post-treatment compared to health and decreased significantly cantly increased at 12-months compared to 3- and 6-months in the at 3- and 12-months compared to baseline. In the moderate moderate TxO group. However, no treatment-related changes were + and poor TxO groups, IFN-γ cells were significantly lower at observed in the poor TxO group.

+ + + + FIGURE 3 IFN-γ and Foxp3 cells in PBMCs. IFN-γ (a) cells and Foxp3 (b) cells as a percentage of PBMCs and separated based on treatment outcome (TxO) groups as good, moderate and poor in health and periodontitis at baseline and 3-, 6- and 12-months. Line graphs display the mean and SD and show differences between the three TxO groups; symbols above the plots represent p ≤ 0.05 between *good- moderate, †moderate-poor, and ‡good-poor TxO groups. Box and whisker plots display the median, interquartile ranges and 10th–90th percentiles and show differences within timepoint or the three TxO groups; above the plots represents p ≤ 0.05. Analysis of significance was calculated using independent samples t test for differences between health and each time point in periodontitis, and linear mixed models for variation with timepoint and treatment outcome in periodontitis. MEDARA et al. | 83 84 | MEDARA et al. MEDARA et al. | 85

FIGURE 4 IFN-γ, IL-4, IL-17 and IL-10 cytokine production in PBMC supernatant with various stimulation methods. IFN-γ (a), IL-4 (b), IL-17 (c) and IL-10 (d) in pg/ml produced by PBMCs without any stimulation, and upon P. gingivalis and concanavalin (ConA) stimulation in health and at baseline, and 3-, 6-, 12-months in periodontitis. Box and whisker plots represent the median, interquartile ranges and 10th–90th percentiles and show differences within timepoint; above the plots represents p ≤ 0.05. Analysis of significance was calculated using independent samples t test for differences between health and each timepoint in periodontitis, and linear mixed models for variation with timepoint in periodontitis.

No definitive patterns were found for total IL-4+ and IL-17+ cells, Likewise, for the base LMMs for P. gingivalis stimulation of PBMCs, + + + + and combinations of CD4 or TCRαβ expression in IFN-γ , IL-4 , mean PD was a significant predictor for IL-17 production (b = 7.82, + + IL-17 and Foxp3 cells or double-positive cytokine expression in SE = 3.00) and %BOP was a significant predictor for IFN-γ produc- + + TCRαβ CD4 cells and the data are presented in Appendix S1, Tables tion (b = 0.42, SE = 0.19) in the model for timepoint. A1 and A2.

4 | DISCUSSION 3.3 | Cytokine production in PBMC supernatant uponP. gingivalisstimulation Many studies have proposed a role for Th1, Th2, Th17 and Treg cells + + + in the pathogenesis of periodontitis based on IFN-γ , IL-4 , IL-17 or Stimulation with ConA elicited the greatest amount of cytokine pro- IL-10 + expression (Chen et al., 2016; Luo et al., 2014). However, these duction, followed by P. gingivalis and cells alone. Upon stimulation studies require careful interpretation as the expression of these cy- + + with P. gingivalis, IL-10 was the largest cytokine produced followed tokines is not restricted to CD4 TCRαβ cells in PBMCs. This study + + + + by IFN-γ (Appendix S1, Figures A2 and A3). found that the majority of Foxp3 and IL-17 cells were TCRαβ CD4 , + + IFN-γ, IL-4, IL-17 and IL-10 production with various stimulation that is true T helper cells, whereas the majority of IFN-γ and IL-4 + − + + methods are displayed in Figure 4. Upon stimulation with P. gingi- cells were TCRαβ CD4 followed by TCRαβ CD4 cells. Likely, these + − + + − valis, IFN-γ was lower at all timepoints, IL-4 was lower at 3-, 6- and TCRαβ CD4 cells are CD8 T cells, CD4 TCRαβ cells are NK and − − 12-months, and IL-17 was lower at baseline and 3-months in com- NKT cells, and TCRαβ CD4 cells are γδ T cells, macrophages, ba- − − parison with health. With treatment, IFN-γ was decreased at 3- and sophils or B cells in PBMCs. Interestingly, TCRαβ CD4 cells were 6-months and IL-4 was decreased at all timepoints compared to not the smallest subset expressing IFN-γ and IL-4, and thus, other baseline, whereas no treatment-related changes were observed for non-canonical cytokine-producing cells may also contribute to pe- IL-17 and IL-10. ripheral immune dysregulation in periodontitis. In line with this, + − + The data for IFN-γ, IL-4, IL-17 and IL-10 production in PBMC su- TCRαβ CD4 FoxP3 cells were significantly positively correlated pernatant when separated based on TxO groups are presented in with health. Previous studies that investigated other cellular sources Appendix S1, Figures A4–A7. of cytokines have reported that IFN-γ mRNA was equally frequently detected in both CD4+ and CD8+ gingival mononuclear cells (GMCs) in periodontitis patients (Takeichi et al., 2000). Likewise, GMCs ex- 3.4 | Associations with periodontal parameters tracted from periodontitis and healthy patients showed no signifi- + + + + cant differences in IFN-γ and IL-4 expression in CD4 and CD8 Tcell phenotypes and cytokine production with various stimuli were cells or IL-10+ in CD4+ cells, however, IL-10+ in CD8+ cells was sig- correlated with mean PD and %BOP and are displayed in Tables 2 nificantly lower in periodontitis compared to health (Gemmell & and 3. Seymour, 1998). + When periodontal parameters were added as fixed effects IFN-γ cells were significantly lower at baseline, and 3- and to the base LMMs for Tcell phenotypes, %BOP was a signifi- 6-months post-treatment compared to health which is contrary to + − + + + cant predictor for TCRαβ CD4 IFN-γ (estimate of fixed effect several studies in peripheral blood that reported higher CD4 IFN-γ (b) = 0.13; standard error (SE) = 0.05 for timepoint, SE = 0.06 cells in periodontitis compared to health (Chen et al., 2016; Schmidt + + + + for TxO), TCRαβ CD4 Foxp3 IL-17 (b = 0.08; SE = 0.02) and et al., 2014) or reported no significant differences (Cheng et al., 2018; + + + + + TCRαβ CD4 CD25 Foxp3 (b = 0.08; SE = 0.03) cells in the models Lima et al., 2011; Okada et al., 2017). Fewer IFN-γ expressing cells for both timepoint and TxO. A 1% increase in BOP was associated at baseline were mirrored by lower IFN-γ production upon stimula- + − + with 0.13% increase in TCRαβ CD4 IFN-γ cells, and 0.08% increase tion with P. gingivalis. One other study reported lower IFN-γ in ear- + + + + + + + + in both TCRαβ CD4 Foxp3 IL-17 and TCRαβ CD4 CD25 Foxp3 ly-onset periodontitis compared to chronic periodontitis and health cells with both timepoint and treatment outcome. Mean PD in PBMC supernatant upon stimulation with phytohaemagglutinin (b = −0.79, SE = 0.37) and %BOP (b = 0.02, SE = 0.01) were both (Sigusch et al., 1998). This depressed IFN-γ response may be a func- + + + + significant predictors for TCRαβ CD4 IL-4 Foxp3 cells in the base tion of compartmentalization of immune system. The periodontal model for treatment outcome. A 1 mm increase in mean PD was sulcus perpetually harbours microbes, and consequently, there is al- associated with 0.79 mm decrease, and a 1% increase in BOP was ways sub-clinical inflammation in the periodontal connective tissue. + + + + associated with 0.02% increase in TCRαβ CD4 IL-4 Foxp3 cells. Thus, compartmentalization within the local and systemic immune 86 | MEDARA et al.

TABLE 2 Correlations between cell phenotypes and periodontal parameters.

Periodontitis

Health Baseline 3-months 6-months 12-months

Mean Mean Mean Mean Mean Cell phenotypes PD %BOP PD %BOP PD %BOP PD %BOP PD %BOP

+ IFN-γ cells (% of PBMCs) 0.011 −0.007 −0.012 0.037 0.148 0.336 0.410 0.571 0.533 0.357 + − + CD4 TCRαβ cells (% IFN-γ 0.246 −0.032 −0.104 −0.125 −0.108 −0.349 0.045 0.053 −0.022 0.025 cells) + + + TCRαβ CD4 cells(% IFN-γ −0.006 −0.070 −0.232 −0.401 −0.051 −0.445 −0.366 −0.297 0.126 −0.231 cells) + − + TCRαβ CD4 cells(% IFN-γ −0.125 0.053 0.319 0.558 0.054 0.383 0.126 0.151 0.053 0.157 cells) − − + TCRαβ CD4 cells(% IFN-γ 0.183 0.098 −0.105 −0.189 0.019 0.141 0.254 0.266 −0.068 0.212 cells) IL−4 + cells (% of PBMCs) −0.005 0.028 −0.034 0.012 0.311 0.140 −0.041 −0.088 0.054 0.195 + − + CD4 TCRαβ cells(% IL−4 0.065 −0.030 0.187 0.019 −0.057 −0.400 −0.157 −0.115 −0.121 −0.136 cells) + + + TCRαβ CD4 cells(% IL−4 0.083 0.052 −0.016 −0.026 −0.194 −0.181 0.021 −0.085 0.051 −0.212 cells) + − + TCRαβ CD4 cells(% IL−4 −0.068 0.064 0.006 0.028 0.183 0.239 0.001 0.132 0.055 0.228 cells) − − + TCRαβ CD4 cells(% IL−4 0.099 −0.071 0.0035 −0.008 −0.048 −0.179 −0.203 −0.175 0.008 0.144 cells) IL−17+ cells (% of PBMCs) 0.014 0.153 0.069 −0.119 0.214 0.271 0.353 0.369 0.417 0.156 + − + CD4 TCRαβ cells(% IL−17 0.026 −0.058 −0.107 −0.043 −0.227 −0.367 −0.158 −0.078 −0.080 0.062 cells) + + + TCRαβ CD4 cells(% IL−17 −0.025 −0.118 0.074 −0.077 0.147 0.331 0.244 0.300 0.017 −0.174 cells) + − + TCRαβ CD4 cells(% IL−17 0.067 0.223 0.013 0.080 −0.133 −0.264 −0.206 −0.229 0.015 0.206 cells) − − + TCRαβ CD4 cells(% IL−17 0.123 0.168 −0.064 0.004 −0.031 −0.108 −0.341 −0.440 −0.017 0.184 cells) Foxp3+ cells (% of PBMCs) −0.054 −0.068 −0.022 −0.127 0.183 0.073 0.071 0.372 −0.037 0.120 + − + CD4 TCRαβ cells(% Foxp3 −0.027 −0.105 0.043 0.031 −0.272 −0.401 −0.056 −0.036 −0.024 0.067 cells) + + + TCRαβ CD4 cells(% Foxp3 −0.220 −0.005 −0.147 −0.278 0.059 0.332 0.262 0.227 −0.057 −0.140 cells) + − + TCRαβ CD4 cells(% Foxp3 0.383 0.159 0.094 0.215 −0.031 −0.262 −0.267 −0.030 0.049 0.093 cells) − − + TCRαβ CD4 cells(% Foxp3 0.010 −0.025 −0.027 0.102 −0.017 −0.184 −0.114 −0.117 −0.011 0.047 cells) + + IFNγ IL−4 cells(% −0.024 0.033 0.103 0.275 −0.002 0.001 0.072 −0.017 −0.083 0.152 + + TCRαβ CD4 cells) + + IFNγ Foxp3 cells(% −0.109 −0.282 0.018 0.160 0.111 0.120 0.153 0.223 −0.316 −0.224 + + TCRαβ CD4 cells) IL−4 +Foxp3+ cells(% −0.248 −0.182 0.129 0.379 −0.104 −0.152 0.062 0.161 −0.161 −0.063 + + TCRαβ CD4 cells) IL−4 +IL−17+ cells(% 0.098 0.326 −0.117 −0.048 0.080 0.066 −0.050 −0.169 −0.066 0.104 + + TCRαβ CD4 cells) + + IFNγ IL−17 cells(% 0.179 0.170 −0.125 −0.298 0.009 −0.064 0.126 0.224 0.253 0.027 + + TCRαβ CD4 cells)

(Continues) MEDARA et al. | 87

TABLE 2 (Continued)

Periodontitis

Health Baseline 3-months 6-months 12-months

Mean Mean Mean Mean Mean Cell phenotypes PD %BOP PD %BOP PD %BOP PD %BOP PD %BOP

Foxp3+IL−17+ cells(% −0.054 −0.320 −0.061 0.237 0.112 0.049 −0.089 0.309 −0.031 0.216 + + TCRαβ CD4 cells) Foxp3+CD25+ cells(% 0.015 0.009 0.081 0.268 −0.037 0.112 0.111 0.050 0.042 0.075 + + TCRαβ CD4 cells)

Note: Spearman's correlation coefficients (ρ) between the periodontal parameters, mean probing depth (PD) and percentage of sites exhibiting + + + + + − + + + − − − bleeding on probing (%BOP), and IFN-γ , IL-4 , IL-17 and Foxp3 cells divided based on CD4 TCRαβ , TCRαβ CD4 , TCRαβ CD4 and TCRαβ CD4 + + + + + + expression as well as Foxp3 , IFN-γ , IL-4 and IL-17 double-positive TCRαβ CD4 cells in health and at baseline, 3-, 6- and 12-months in periodontitis. Significant correlations (p ≤ 0.05) are represented in bold.

component is an essential survival strategy. Indeed, a previous study other cross-sectional studies that reported no significant differences + + + + + + that explored cytokine mRNA expression in gingival tissues and in CD4 IL-17 , CD8 IL-17 or TCRγδ IL-17 cells between health and PBMCs reported higher IFN-γ in PBMCs compared to gingival tis- periodontitis (Cheng et al., 2018; Okada et al., 2017). Nonetheless, sues and higher IL-10 in gingival tissues compared to peripheral blood this does not exclude a role for Th2 (Aoyagi et al., 1995) and Th17 (Yamazaki et al., 1997). The depressed IFN-γ response in periodon- cells (Cardoso et al., 2009) at the gingival interface. titis might then indicate a hyper-responsive suppression of normal ConA stimulation of PBMCs elicited greater cytokine production immune function. Thus, paradoxically, although there were lower compared to PBMCs alone in line with previous studies (Berker et al., + IFN-γ expressing cells in periodontitis in the present study, variation 2013; Fujihashi et al., 1993; Gonçalves et al., 2013; McFarlane et al., with treatment was observed in the good TxO group, whereas no 1990). Upon P. gingivalis stimulation, IL-10 was the most abundant + + changes with treatment were observed in IFN-γ and Foxp3 cells cytokine produced with 3.5-fold and seven-fold higher mean con- in the poor TxO group. This decrease with treatment may represent centration than IFN-γ at health and baseline, respectively. IL-10 has a reduction in the overall inflammatory burden representing good many anti-inflammatory functions including downregulating MHC health outcomes. Accordingly, mean PD and %BOP were signifi- class ΙΙ and co-stimulatory molecule expression on APCs impair- + − + cantly positively correlated with TCRαβ CD4 IFN-γ cells at base- ing their antigen-presenting capacity (de Waal Malefyt et al., 1991; line. Compartmentalization of the immune response has also been Ding et al., 1993; Willems et al., 1994), and negatively regulating shown for other disease states such as sarcoidosis (Hudspith et al., Th1, Th2 and Th17 cells (Coomes et al., 2017; O'Garra & Vieira, 1987) and tuberculosis (Guglielmetti et al., 2013). 2007) while sustaining Treg cell production (Murai et al., 2009). Treg cells are involved in the resolution of inflammation and Thus, it follows that incubation with periodontal pathogens would maintaining homeostatic control of immune responses. Their prin- elicit large amounts of IL-10 production irrespective of health or cipal suppressive role is mainly via inhibition of APC maturation and disease. This study, like others (Borch et al., 2010; de Heens et al., effector cell differentiation, and secretion of inhibitory cytokines, 2009; Goncalves et al., 2010), found no differences in IL-10 between such as IL-10, TGF-β and IL-35 (Sakaguchi et al., 2010). It has also periodontitis and health, however, others reported higher produc- been shown that during disease resolution, T cells change their tran- tion in periodontitis supporting an anti-inflammatory role for IL-10 scriptional profile and acquire regulatory properties in the presence (Gonçalves et al., 2013; Trindade et al., 2012). P. gingivalis stimulation + of TGF-β (Gagliani et al., 2015). Thus, the increase in Foxp3 cells at only elicited low amounts of IL-4 (timepoint total mean: 2.93 pg/ml; 12-months seen in this study may be associated with the conversion range 0.00–18.82 pg/ml) and IL-17 (timepoint total mean: 7.18 pg/ of inflammatory cells into regulatory cells and the resolution of in- ml; range 0.00–155.54 pg/ml). Moreover, at 6months, there was no flammation. The late increase in Foxp3+ cells at 12-months rather statistically significant difference in IL-17 produced with and without than early increase corresponding to treatment may reflect Treg cell P. gingivalis stimulation. Low levels of IL-4 and IL-17 with antigens role in periodontal tissue repair and remodelling which may last for a and not common mitogens are also seen in other studies (Kurtzhals year or more after treatment (Gurtner et al., 2008; Waerhaug, 1978; et al., 1992; Lenarczyk et al., 2000). Apart from IL-10–mediated sup- Wilson et al., 2008). pression, other potential reasons for low IL-4 and IL-17 production There were neither any significant differences in total IL-17+ or include short half-life after secretion, rapid degradation of the ex- + + + IL-4 cell proportions nor when separated into CD4 and TCRαβ pressed cytokine or possible sequestration of the cytokine in cul- subsets. During data collection with flow cytometry, the mean num- ture by receptor on other leucocytes. Moreover, activation of αβ T + + ber IL-17 and IL-4 cells acquired within the cytokine gate was 1,900 cells, γδ cells or NK cells may not be directly mediated by bacterial and 4,500 cells, respectively, which is well above the threshold of products, and co-stimulatory molecules like CD28, effective antigen 100 cells for rare populations (Roederer, 2008). This corroborates processing by APCs, and the strength and duration of TCR signalling 88 | MEDARA et al.

TABLE 3 Correlations between cytokine production in PBMC supernatant upon various stimulation methods and periodontal parameters.

Periodontitis

Health Baseline 3-months 6-months 12-months

Mean Mean Mean Mean Mean Cytokines PD %BOP PD %BOP PD %BOP PD %BOP PD %BOP

IFN-γ -cells alone 0.222 0.081 0.034 −0.100 0.302 0.305 0.542 0.493 −0.116 −0.216 IFN-γ- P. gingivalis 0.021 −0.010 0.106 0.158 −0.004 0.302 0.305 0.442 0.030 −0.115 stimulation

IFN-γ-ConA stimulation 0.053 −0.141 −0.090 −0.075 −0.297 −0.284 −0.106 0.183 0.192 0.191 IL−4-cells alone 0.124 −0.141 −0.062 0.080 −0.181 −0.243 0.027 −0.144 −0.356 −0.412 IL−4- P. gingivalis 0.225 −0.079 −0.104 0.077 −0.253 −0.171 0.091 −0.082 −0.367 −0.451 stimulation IL−4-ConA stimulation 0.024 −0.339 −0.035 −0.105 0.345 0.290 0.499 0.167 0.119 0.041 IL−17-cells alone 0.077 −0.001 −0.041 −0.290 0.316 0.206 0.171 0.135 −0.149 −0.328 IL−17- P. gingivalis 0.031 0.043 0.272 0.099 0.177 0.425 0.578 0.490 −0.109 −0.292 stimulation IL−17-ConA stimulation −0.128 −0.103 0.329 0.021 0.296 0.499 0.441 0.337 0.287 0.266 IL−10-cells alone −0.066 0.272 0.223 0.122 0.244 −0.142 −0.124 −0.047 −0.065 −0.210 IL−10 - P. gingivalis −0.112 −0.171 0.065 0.194 0.088 −0.062 0.004 −0.101 0.338 0.429 stimulation IL−10-ConA stimulation −0.052 −0.140 0.131 −0.077 0.176 0.326 0.311 0.364 −0.033 −0.198

Note: Spearman's correlation coefficients (ρ) between the periodontal parameters, mean probing depth (PD) and percentage of sites exhibiting bleeding on probing (%BOP), and IFN-γ, IL-4, IL-17 and IL-10 cytokine expression from PBMCs without any stimulation and upon P. gingivalis and concanavalin A (ConA) stimulation in health and at baseline, 3-, 6- and 12-months in periodontitis. Significant correlations (p ≤ 0.05) are represented in bold. may all affect efficient T-cell activation (Champaiboon et al., 2000; et al., 2010), and various preparation methods such as forma- Gemmell & Seymour, 2004). lin-killed whole cells and sonicates (Aoyagi et al., 1995) or virulence + − + − In this study, we can only speculate CD4 TCRαβ , TCRαβ CD4 factors (Trindade et al., 2012). Mean PD and %BOP were the sole − − or CD4 TCRαβ cell lineages and further studies with specific mark- clinical periodontal parameters used in this study as the treatment ers that identify other cellular sources in PBMCs would provide a goal was shallow pockets that do not bleed when probed. Other pa- broader understanding of the host immune response in periodonti- rameters such as clinical attachment levels and radiographic bone tis. The T helper cell family has also expanded to include Th9, Th22 loss may provide a more dynamic picture of the periodontal status and T follicular helper cells and longitudinal clinical studies in this and peripheral immune response of the study cohort. Lastly, the ini- area are also needed. T cells typically demonstrate low to unde- tial study parameters were set in 2013 well before the definition tectable amounts of spontaneous cytokine expression requiring in of health was established in the current periodontal classification. vitro activation using antigenic, mitogenic or pharmacologic (PMA/ Thus, the healthy control group in this study includes health and lo- ionomycin) stimuli. Previous studies have reported CD4 downreg- calized gingivitis cases and the results should be interpreted in light ulation upon stimulation with PMA/ionomycin (O'Neil-Andersen & of this (Cheng et al., 2018). + Lawrence, 2002), and higher IFN-γ and IL-17 secretion in CD4 Tcell Th cells play a key role in periodontitis immunopathology, and clones with PMA/ionomycin stimulation compared to anti-CD3/ the contribution of Th cells to this process remains unresolved. CD28 (Olsen & Sollid, 2013). Likewise in PBMC supernatants, this This study is the first to report longitudinal variation in four ca- study did not identify which specific cells contributed to the cyto- nonical cytokines for Th1, Th2, Th17 and Treg cells in response kine milieu. Kobayashi et al. (2000) have reported 30-fold higher to P. gingivalis challenge in PBMCs. In cell supernatant, IFN-γ and IFN-γ production by PBMCs compared to purified T cells when stim- IL-17 were significantly lower at baseline compared to health, ulated with Aggregatibacter actinomycetemcomitans, suggesting that and treatment resulted in a reduction in IFN-γ and IL-4 compared + + T cells may not be the major cytokine contributors. This study used to baseline. Moreover, this study found that TCRαβ CD4 and + − + + whole-cell lysate which represents the entire cohort of virulence TCRαβ CD4 cells were the major sources of IFN-γ and IL-4 + + factors that could elicit an immune response. Previous literature expression in PBMCs whereas TCRαβ CD4 cells were the major has reported differential cytokine responses with different bacte- sources of Foxp3+ and IL-17+ cells. The good TxO group showed + rial species (Kobayashi et al., 2000), live and dead bacteria (Scheres a treatment-related reduction in IFN-γ , whereas no changes in MEDARA et al. | 89

+ + Byrne, S., Dashper, S., Darby, I., Adams, G., Hoffmann, B., & Reynolds, E. IFN-γ or Foxp3 cells were observed in the poor TxO group. (2009). Progression of chronic periodontitis can be predicted by the Collectively, this study corroborates the presence of distinct Th levels of Porphyromonas gingivalis and Treponema Denticola in sub- cell profiles in peripheral blood during management of periodonti- gingival plaque. Oral Microbiology and Immunology, 24(6), 469–477. tis and highlights the complexity of systemic immune responses at Cardoso, C., Garlet, G., Crippa, G., Rosa, A., Junior, W., Rossi, M., & Silva, sites distant to the local bacterial challenge. J. (2009). Evidence of the presence of t helper type 17 cells in chronic lesions of human periodontal disease. Oral Microbiology and Immunology, 24(1), 1–6. ACKNOWLEDGMENT Champaiboon, C., Yongvanitchit, K., Pichyangkul, S., & Mahanonda, R. The authors wish to acknowledge the Melbourne Cytometry (2000). The immune modulation of b-cell responses by porphyro- Platform for equipment use and expertise. monas gingivalis and interleukin-10. Journal of Periodontology, 71(3), The Australian National Health and Medical Research Council 468–475. Chapple, I. L., & Genco, R. (2013). Diabetes and periodontal diseases: (NHMRC) of Australia and Australian Research Council (ARC) are Consensus report of the joint EFP/AAP workshop on periodonti- thanked for financial support over many years for the immunology tis and systemic diseases. Journal of Periodontology, 84(4 Suppl), studies reported in the authors’ laboratories.NMOBS is the recipient S106–112. of NHMRC funding (APP1142472, APP1158841, APP1185426), ARC Chen, X.-T., Chen, L.-L., Tan, J.-Y., Shi, D.-H., Ke, T., & Lei, L.-H. (2016). Th17 and Th1 lymphocytes are correlated with chronic periodonti- funding (DP160101312, LE200100163), Cancer Council Victoria tis. Immunological Investigations, 45(3), 243–254. funding (APP1163284) and ADRF in immunology and research is Cheng, W. C., Saleh, F., Abuaisha Karim, B., Hughes, F. J., & Taams, L. S. supported by the Centre for Oral Health Research at The Melbourne (2018). Comparative analysis of immune cell subsets in peripheral Dental School. NM was supported by the Oral Health Cooperative blood from patients with periodontal disease and healthy controls. Clinical and Experimental Immunology, 194(3), 380–390. Research Centre Top-up Scholarship. Coomes, S. M., Kannan, Y., Pelly, V. S., Entwistle, L. J., Guidi, R., Perez- Lloret, J., Nikolov, N., Muller, W., & Wilson, M. S. (2017). CD4(+) Th2 CONFLICT OF INTEREST cells are directly regulated by IL-10 during allergic airway inflamma- The authors have stated explicitly that there are no conflicts of inter- tion. Mucosal Immunology, 10(1), 150–161. https://doi.org/10.1038/ mi.2016.47 est in connection with this article. Cooney, L. A., Towery, K., Endres, J., & Fox, D. A. (2011). Sensitivity and resistance to regulation by IL-4 during Th17 maturation. The Journal ORCID of Immunology, 187(9), 4440–4450. Nidhi Medara https://orcid.org/0000-0002-5450-0647 Darby, I. B., Mooney, J., & Kinane, D. F. (2001). Changes in subgingival microflora and humoral immune response following periodontal Eric C. Reynolds https://orcid.org/0000-0002-6618-4856 therapy. Journal of Clinical Periodontology, 28(8), 796–805. https:// Ivan B. Darby https://orcid.org/0000-0002-6457-5327 doi.org/10.1034/j.1600-051X.2001.280812.x de Heens, G. L., Kikkert, R., Aarden, L. A., van der Velden, U., & Loos, B. REFERENCES G. (2009). Effects of smoking on the ex vivo cytokine production in periodontitis. , (1), 28–34. Aoyagi, T., Sugawara-Aoyagi, M., Yamazaki, K., & Hara, K. (1995). Journal of Periodontal Research 44 de Waal, M. R., Haanen, J., Spits, H., Roncarolo, M. G., te Velde, A., Interleukin 4 (IL-4) and IL-6-producing memory T-cells in peripheral Figdor, C., Johnson, K., Kastelein, R., Yssel, H., & de Vries, J. E. blood and gingival tissues in periodontitis patients with high serum (1991). Interleukin 10 (IL-10) and viral IL-10 strongly reduce an- antibody titers to Porphyromonas Gingivalis. Oral Microbiology and tigen-specific human T cell proliferation by diminishing the an- Immunology, 10(5), 304–310. tigen-presenting capacity of monocytes via downregulation of Berker, E., Kantarci, A., Hasturk, H., & Van Dyke, T. E. (2013). Blocking class ii major histocompatibility complex expression. proinflammatory cytokine release modulates peripheral blood Journal of Experimental Medicine, 174(4), 915–924. https://doi.org/10.1084/ mononuclear cell response to Porphyromonas Gingivalis. Journal of jem.174.4.915 Periodontology, 84(9), 1337–1345. Dietrich, T., Sharma, P., Walter, C., Weston, P., & Beck, J. (2013). The ep- Blasco-Baque, V., Garidou, L., Pomié, C., Escoula, Q., Loubieres, P., Le idemiological evidence behind the association between periodon- Gall-David, S., Lemaitre, M., Nicolas, S., Klopp, P., Waget, A., titis and incident atherosclerotic cardiovascular disease. Journal Azalbert, V. (2017). Periodontitis induced by Porphyromonas gingi- of Clinical Periodontology, 40(Suppl 14), S70–S84. https://doi. valis drives periodontal microbiota dysbiosis and insulin resistance org/10.1111/jcpe.12062 via an impaired adaptive immune response. Gut, 66(5), 872–885. Ding, L., Linsley, P. S., Huang, L. Y., Germain, R. N., & Shevach, E. M. Borch, T. S., Holmstrup, P., Bendtzen, K., & Nielsen, C. H. (2010). In vitro (1993). IL-10 inhibits macrophage costimulatory activity by selec- cytokine responses to periodontal pathogens: Generalized aggres- tively inhibiting the up-regulation of B7 expression. sive periodontitis is associated with increased IL-6 response to The Journal of Immunology, 151(3), 1224–1234. Porphyromonas gingivalis. Scandinavian Journal of Immunology, 71(6), Fujihashi, K., Beagley, K. W., Kono, Y., Aicher, W. K., Yamamoto, M., 440–446. DiFabio, S., Xu- Amano, J., McGhee, J. R., & Kiyono, H. (1993). Borgnakke, W. S., Ylöstalo, P. V., Taylor, G. W., & Genco, R. J. (2013). Gingival mononuclear cells from chronic inflammatory periodon- Effect of periodontal disease on diabetes: Systematic review of ep- tal tissues produce interleukin (IL)-5 and IL-6 but not IL-2 and IL-4. idemiologic observational evidence. Journal of Periodontology, 84(4 , (4), 1239–1250. Suppl), S135–S152. American Journal of Pathology 142 Gagliani, N., Vesely, M. C. A., Iseppon, A., Brockmann, L., Xu, H., Palm, N. Bui, F. Q., Almeida-da-Silva, C. L. C., Huynh, B., Trinh, A., Liu, J., W., de Zoete, M. R., Licona-Limon, P., Paiva, R. S., Ching, T., Weaver, Woodward, J., Asadi, H., & Ojcius, D. M. (2019). Association be- C. (2015). Th17 cells transdifferentiate into regulatory T cells during tween periodontal pathogens and systemic disease. Biomedical resolution of inflammation. Nature, 523(7559), 221–225. Journal, 42(1), 27–35. 90 | MEDARA et al.

Gemmell, E., & Seymour, G. J. (1998). Cytokine profiles of cells extracted Journal of Immunological Methods, 156(2), 239–245. https://doi. from humans with periodontal diseases. Journal of Dental Research, org/10.1016/0022-1759(92)90031-N 77(1), 16–26. https://doi.org/10.1177/00220345980770010101 Lalla, E., Kaplan, S., Yang, J., Roth, G. A., Papapanou, P. N., & Greenberg, Gemmell, E., & Seymour, G. J. (2004). Immunoregulatory control of Th1/ S. (2007). Effects of periodontal therapy on serum C-reactive pro- Th2 cytokine profiles in periodontal disease. Periodontol, 2000(35), tein, se-selectin, and tumor necrosis factor-α secretion by periph- 21–41. https://doi.org/10.1111/j.0906-6713.2004.003557.x eral blood-derived macrophages in diabetes. A Pilot Study. Journal Gonçalves, P. F., Klepac-Ceraj, V., Huang, H., Paster, B. J., Aukhil, I., of Periodontal Research, 42(3), 274–282. Wallet, S. M., & Shaddox, L. M. (2013). Correlation of aggregati- Lam, R. S., O'Brien-Simpson, N. M., Hamilton, J. A., Lenzo, J. C., Holden, bacter actinomycetemcomitans detection with clinical/immunoin- J. A., Brammar, G. C., Orth, R. K., Tan, Y., Walsh, K. A., Fleetwood, flammatory profile of localized aggressive periodontitis using a 16s A. J., Reynolds, E. C. (2015). GM-CSF and uPA are required for rRNA microarray method: A cross-sectional study. PLoS One, 8(12), Porphyromonas Gingivalis-induced alveolar bone loss in a mouse e85066. https://doi.org/10.1371/journal.pone.0085066 periodontitis model. Immunology and Cell Biology, 93(8), 705–715. Goncalves, T. O., Costa, D., Brodskyn, C. I., Duarte, P. M., Cesar Neto, Lenarczyk, A., Helsloot, J., Farmer, K., Peters, L., Sturgess, A., & Kirkham, J. B., & Nogueira-Filho, G. (2010). Release of cytokines by stimu- B. (2000). Antigen-induced IL-17 response in the peripheral lated peripheral blood mononuclear cells in chronic periodontitis. blood mononuclear cells (PBMC) of healthy controls. Clinical and Archives of Oral Biology, 55(12), 975–980. https://doi.org/10.1016/j. Experimental Immunology, 122(1), 41–48. archoralbio.2010.08.002 Lima, P. M., Souza, P. E., Costa, J. E., Gomez, R. S., Gollob, K. J., & Dutra, Grünig, G., Corry, D. B., Leach, M. W., Seymour, B. W. P., Kurup, V. P., W. O. (2011). Aggressive and chronic periodontitis correlate with & Rennick, D. M. (1997). Interleukin-10 is a natural suppressor of distinct cellular sources of key immunoregulatory cytokines. cytokine production and inflammation in a murine model of allergic Journal of Periodontology, 82(1), 86–95. bronchopulmonary aspergillosis. Journal of Experimental Medicine, Luo, Z., Wang, H., Wu, Y., Sun, Z., & Wu, Y. (2014). Clinical significance of 185(6), 1089–1100. https://doi.org/10.1084/jem.185.6.1089 IL-23 regulating IL-17A and/or IL-17F positive Th17 cells in chronic Guglielmetti, L., Cazzadori, A., Conti, M., Boccafoglio, F., Vella, A., periodontitis. Mediators of Inflammation, 2014, 627959. Ortolani, R., & Concia, E. (2013). Lymphocyte subpopulations in McFarlane, C. G., Reynolds, J. J., & Meikle, M. C. (1990). The release active tuberculosis: Association with disease severity and the QFT- of interleukin-1 beta, tumor necrosis factor-alpha and interfer- GIT assay. The International Journal of Tuberculosis and Lung Disease, on-gamma by cultured peripheral blood mononuclear cells from 17(6), 825–828. patients with periodontitis. Journal of Periodontal Research, 25(4), Gurtner, G. C., Werner, S., Barrandon, Y., & Longaker, M. T. (2008). 207–214. Wound repair and regeneration. Nature, 453(7193), 314–321. Mosmann, T. R., Cherwinski, H., Bond, M. W., Giedlin, M. A., & Coffman, Hajishengallis, G., & Korostoff, J. M. (2017). Revisiting the page & R. L. (1986). Two types of murine helper T cell clone. I. Definition Schroeder model: The good, the bad and the unknowns in the according to profiles of lymphokine activities and secreted pro- periodontal host response 40 years later. Periodontology, 75(1), teins. The Journal of Immunology, 136(7), 2348–2357. 116–151. Murai, M., Turovskaya, O., Kim, G., Madan, R., Karp, C. L., Cheroutre, H., Hajishengallis, G., Liang, S., Payne, M. A., Hashim, A., Jotwani, R., Eskan, & Kronenberg, M. (2009). Interleukin 10 acts on regulatory T cells M. A., McIntosh, M. L., Alsam, A., Kirkwood, K. L., Lambris, J. D., to maintain expression of the transcription factor FOXP3 and sup- Darveau, R. P., & Curtis, M. A. (2011). Low-abundance biofilm spe- pressive function in mice with colitis. Nature Immunology, 10(11), cies orchestrates inflammatory periodontal disease through the 1178–1184. commensal microbiota and complement. Cell Host & Microbe, 10(5), O'Garra, A., & Vieira, P. (2007). T(h)1 cells control themselves by produc- 497–506. https://doi.org/10.1016/j.chom.2011.10.006 ing interleukin-10. Nature Reviews Immunology, 7(6), 425–428. Harrington, L. E., Hatton, R. D., Mangan, P. R., Turner, H., Murphy, T. L., Okada, K., Fujimura, T., Kikuchi, T., Aino, M., Kamiya, Y., Izawa, A., Murphy, K. M., & Weaver, C. T. (2005). Interleukin 17–producing Iwamura, Y., Goto, H., Okabe, I., Miyake, E., Hasegawa, Y. (2017). CD4+ effector T cells develop via a lineage distinct from the T Effect of interleukin (IL)-35 on IL-17 expression and production by helper type 1 and 2 lineages. Nature Immunology, 6(11), 1123–1132. human CD4(+) T cells. PeerJ, 5, e2999. https://doi.org/10.1038/ni1254 Olsen, I., & Sollid, L. M. (2013). Pitfalls in determining the cytokine pro- Hudspith, B. N., Flint, K. C., Geraint-James, D., Brostoff, J., & file of human T cells. Journal of Immunological Methods, 390(1–2), Johnson, N. M. (1987). Lack of immune deficiency in sarcoidosis: 106–112. Compartmentalisation of the immune response. Thorax, 42(4), 250– O'Neil-Andersen, N. J., & Lawrence, D. A. (2002). Differential modula- 255. https://doi.org/10.1136/thx.42.4.250 tion of surface and intracellular protein expression by T cells after Jain, A., Batista, E. L. Jr, Serhan, C., Stahl, G. L., & Van Dyke, T. E. (2003). stimulation in the presence of monensin or brefeldin A. Clinical and Role for periodontitis in the progression of lipid deposition in an Diagnostic Laboratory Immunology, 9(2), 243–250. animal model. Infection and Immunity, 71(10), 6012–6018. https:// Roederer, M. (2008). How many events is enough? Are you positive? doi.org/10.1128/IAI.71.10.6012-6018.2003 Cytometry Part A, 73A(5), 384–385. Kobayashi, H., Nagasawa, T., Aramaki, M., Mahanonda, R., & Ishikawa, Sabarish, R., Rao, S. R., & Lavu, V. (2016). Natural T regulatory cells I. (2000). Individual diversities in interferon gamma production by (n Treg) in the peripheral blood of healthy subjects and subjects human peripheral blood mononuclear cells stimulated with peri- with chronic periodontitis – A pilot study. Journal of Clinical and odontopathic bacteria. Journal of Periodontal Research, 35(6), 319– Diagnostic Research, 10(3), Zc36–Zc39. 328. https://doi.org/10.1034/j.1600-0765.2000.035006319.x Sakaguchi, S., Miyara, M., Costantino, C. M., & Hafler, D. A. (2010). Konkel, J. E., O'Boyle, C., & Krishnan, S. (2019). Distal consequences of FOXP3+ regulatory T cells in the human immune system. Nature oral inflammation. Frontiers in Immunology, 10(1403). https://doi. Reviews Immunology, 10(7), 490–500. https://doi.org/10.1038/ org/10.3389/fimmu.2019.01403 nri2785 Kurtzhals, J. A., Hansen, M. B., Hey, A. S., & Poulsen, L. K. (1992). Scheres, N., Laine, M. L., de Vries, T. J., Everts, V., & van Winkelhoff, A. J. Measurement of antigen-dependent interleukin-4 production by (2010). Gingival and periodontal ligament fibroblasts differ in their human peripheral blood mononuclear cells. Introduction of an inflammatory response to viable Porphyromonas gingivalis. Journal amplification step using ionomycin and phorbol myristate acetate. of Periodontal Research, 45(2), 262–270. MEDARA et al. | 91

Schmidt, J., Jentsch, H., Stingu, C. S., & Sack, U. (2014). General immune Wilson, T. G. Jr, Carnio, J., Schenk, R., & Myers, G. (2008). Absence of status and oral microbiology in patients with different forms of peri- histologic signs of chronic inflammation following closed subgingi- odontitis and healthy control subjects. PLoS One, 9(10), e109187. val scaling and root planing using the dental endoscope: Human bi- Sigusch, B., Klinger, G., Glockmann, E., & Simon, H. U. (1998). Early-onset opsies – A pilot study. Journal of Periodontology, 79(11), 2036–2041. and adult periodontitis associated with abnormal cytokine produc- https://doi.org/10.1902/jop.2008.080190 tion by activated T lymphocytes. Journal of Periodontology, 69(10), Yamazaki, K., Nakajima, T., Kubota, Y., Gemmell, E., Seymour, G. J., & 1098–1104. Hara, K. (1997). Cytokine messenger RNA expression in chronic in- Takeichi, O., Haber, J., Kawai, T., Smith, D. J., Moro, I., & Taubman, M. flammatory periodontal disease. Oral Microbiology and Immunology, A., (2000). Cytokine profiles of T-lymphocytes from gingival tis- 12(5), 281–287. https://doi.org/10.1111/j.1399-302X.1997.tb003 sues with pathological pocketing. Journal of Dental Research, 79(8), 92.x 1548–1555. Zhao, L., Zhou, Y., Xu, Y., Sun, Y., Li, L., & Chen, W. (2011). Effect of Toh, M. L., Kawashima, M., Zrioual, S., Hot, A., Miossec, P., & Miossec, non-surgical periodontal therapy on the levels of Th17/Th1/Th2 P. (2009). IL-17 inhibits human Th1 differentiation through IL-12R cytokines and their transcription factors in Chinese chronic peri- beta 2 downregulation. Cytokine, 48(3), 226–230. odontitis patients. Journal of Clinical Periodontology, 38(6), 509–516. Tonetti, M. S., & Van Dyke, T. E. (2013). Periodontitis and atheroscle- https://doi.org/10.1111/j.1600-051X.2011.01712.x rotic cardiovascular disease: Consensus report of the joint EFP/ AAP workshop on periodontitis and systemic diseases. Journal of Periodontology, 84(4 Suppl), S24–S29. https://doi.org/10.1902/ SUPPORTING INFORMATION jop.2013.1340019 Trindade, S. C., Olczak, T., Gomes-Filho, I. S., Moura-Costa, L. F., Additional supporting information may be found online in the Cerqueira, E. M., Galdino-Neto, M., Alves, H., Carvalho-Filho, P. C., Supporting Information section. Xavier, M. T., & Meyer, R. (2012). Induction of interleukin (IL)-1beta, IL-10, IL-8 and immunoglobulin G by Porphyromonas gingivalis HmuY in humans. Journal of Periodontal Research, 47(1), 27–32. How to cite this article: Medara N, Lenzo JC, Walsh KA, Waerhaug, J. (1978). Healing of the dento-epithelial junction following subgingival plaque control. I. As observed in human biopsy mate- O’Brien-Simpson NM, Reynolds EC, Darby IB. Peripheral T rial. Journal of Periodontology, 49(1), 1–8. https://doi.org/10.1902/ helper cell profiles during management of periodontitis. J Clin jop.1978.49.1.1 Periodontol. 2021;48:77–91. https://doi.org/10.1111/ Willems, F., Marchant, A., Delville, J. P., Gerard, C., Delvaux, A., Velu, jcpe.13389 T., de Boer, M., & Goldman, M. (1994). Interleukin-10 inhibits B7 and intercellular adhesion molecule-1 expression on human monocytes. European Journal of Immunology, 24(4), 1007–1009. https:// doi.org/10.1002/eji.1830240435 Received: 29 June 2020 | Revised: 29 September 2020 | Accepted: 2 October 2020 DOI: 10.1111/jcpe.13385

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

Periodontal surgery using rhFGF-2 with deproteinized bovine bone mineral or rhFGF-2 alone: 2-year follow-up of a randomized controlled trial

1Department of Periodontology, Tokyo Dental College, Tokyo, Japan Abstract 2Oral Health Science Center, Tokyo Dental Aim: To compare outcomes of rhFGF-2 + DBBM therapy with rhFGF-2 alone in the College, Tokyo, Japan treatment of intrabony defects. This study provides 2-year follow-up results from the 3Tokyo Dental College Chiba Dental Center, Chiba, Japan previous randomized controlled trial. 4Department of Dental Hygiene, Tokyo Materials and Methods: Defects were randomly allocated to receive rhFGF-2 + DBBM Dental Junior College, Tokyo, Japan (test) or rhFGF-2 (control). Treated sites were re-evaluated at 2 years postoperatively,

Correspondence using original clinical and patient-centred measures. Atsushi Saito, Department of Results: Thirty-eight sites were available for re-evaluation. At 2 years, both groups Periodontology, Tokyo Dental College, 2-9-18 Kanda-Misakicho, Chiyoda-ku, showed a significant improvement in clinical attachment level (CAL) from baseline. A Tokyo 101-0061, Japan. gain in CAL of 3.4 ± 1.3 mm in the test group and 3.1 ± 1.5 mm in the control group Email: [email protected] was found. No significant inter-group difference was noted. Both groups showed a Funding information Supported by a grant from the progressive increase in radiographic bone fill (RBF). The test treatment yielded greater Multidisciplinary Research Center for Jaw RBF (56%) compared with the control group (41%). The control treatment performed Disease (MRCJD), Tokyo Dental College (a MEXT Private University Research better in contained defects in terms of CAL and RBF. There was no significant differ- Branding Project) and an Osteology ence in patient-reported outcomes between groups. Foundation Advanced Researcher Grant (No.17-136). Conclusions: At 2-year follow-up, the test and cotrol treatments were similarly effective in improving CAL, whereas the test treatment achieved a significantly greater RBF. In both treatments, favourable clinical, radiographic, and patient-reported outcomes can be sustained for at least 2 years. Trial registration: The University Hospital Medical Information Network-Clinical Trials Registry (UMIN-CTR) 000025257.

KEYWORDS bone graft, deproteinized bovine bone mineral (DBBM), FGF-2, intrabony defects, patient- reported outcome, periodontal regenerative therapy, periodontitis

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2020 The Authors. Journal of Clinical Periodontology published by John Wiley & Sons Ltd

92 | wileyonlinelibrary.com/journal/jcpe J Clin Periodontol. 2021;48:92–100. AOKI et al. | 93

1 | INTRODUCTION Clinical Relevance In the treatment of periodontitis, the regeneration of lost tissues Scientific rationale for the study: It is important to evaluate is one ultimate goal. Signalling molecules such as growth factors the longevity of effects of combination periodontal regener- play a critical role in periodontal regeneration (Lin et al., 2015; ative therapy. This follow-up study aimed to evaluate 2-year Murakami, 2011). Recombinant human platelet-derived growth outcomes of the combination therapy using rhFGF-2 with factor (rhPDGF)-BB and enamel matrix derivative (EMD) have DBBM in the treatment of intrabony periodontal defects. been clinically used with various degrees of success (Khoshkam Principal findings: At 2-year follow-up, no significant differ- et al., 2015; Li et al., 2017; Lin et al., 2015; Sculean et al., 2015). ence in clinical attachment level gains was noted between Among the biological agents, basic fibroblast growth factor (FGF- the use of rhFGF-2 alone and rhFGF-2 + DBBM. At any 2) has received particular interests, due to its potent ability to time point, the combination therapy showed enhanced ra- induce proliferation and angiogenesis in undifferentiated cells diographic bone fill. There was no significant difference in (Murakami, 2011). Based on the large body of evidence from basic scores on oral health-related quality of life between groups. research, extensive clinical trials were conducted in Japan, and Practical implications: With both surgical interventions, the results indicated that the use of recombinant human FGF-2 favourable clinical and patient-reported outcomes can be (rhFGF-2) in surgical periodontal treatment is safe and clinically sustained for at least 2 years. effective (Kitamura et al., 2008, 2011, 2016; Murakami, 2011). Since 2016, rhFGF-2 has been used as a novel periodontal regenerative therapy in Japan (Saito et al., 2019). In some cases, such as those with non-contained intrabony 2 | MATERIALS AND METHODS defects, the use of biological agents alone may not be sufficient, due to their inability to maintain appropriate space for periodon- 2.1 | Study design tal regeneration (Iorio-Siciliano et al., 2011; Reynolds et al., 2015). To further expand their clinical applicability, the effectiveness of A 2-year follow-up was undertaken in an RCT (Saito et al., 2019) con- combination therapy with bone graft materials has been explored ducted at Tokyo Dental College Suidobashi Hospital (Tokyo, Japan) (Hoffmann et al., 2016; Iorio-Siciliano et al., 2014; Matarasso et al., and Chiba Dental Center (Chiba, Japan), which had involved evalu- 2015; Nevins et al., 2013; Zucchelli et al., 2003). In a systematic ation of healing at up to 6 months postoperatively. In the original review, it has been reported that EMD plus bone grafts may yield study, a two-centre, single-blind, randomized, controlled design was enhanced clinical outcomes compared to the use of EMD alone used. The study was conducted in accordance with the Declaration (Matarasso et al., 2015). of Helsinki and approved by the institutional review board at Tokyo Deproteinized bovine bone mineral (DBBM) has been widely Dental College (No. 747). used as a scaffold material in periodontal treatment (Camargo et al., 2000; Stavropoulos & Karring, 2010). It is often used with a collagen barrier as a guided tissue regeneration (GTR) method. Studies by our 2.2 | Participants group and others showed clinical effectiveness of the combination GTR (Irokawa, Okubo, et al., 2017; Irokawa, Takeuchi, et al., 2017; In the previous RCT (Saito et al., 2019), 32 patients with moderate Sculean et al., 2008; Stavropoulos & Karring, 2010; Tonetti et al., to severe chronic periodontitis (Armitage, 1999; Page & Eke, 2007) 2004). However, no information was available how the combined use were included (recruited between January 2017 and February of rhFGF-2 with DBBM performs in periodontal healing. We hypothe- 2018). A total of 44 intrabony defects were randomized into two sized that enhanced regenerative outcome would be achieved by such groups (Figure S1 flowchart). combination therapy. Therefore, we set out to conduct a randomized The initial inclusion criteria were as follows: presence of an in- controlled trial (RCT) comparing the use of rhFGF-2 plus DBBM and trabony defect depth of ≥3 mm in inter-proximal areas of teeth, sites rhFGF-2 alone to treat intrabony defects (Saito et al., 2019). In this with probing pocket depth (PPD) ≥4 mm, and an adequate level of earlier RCT, we reported that both treatments yielded significant plaque control. Detailed information on the inclusion and exclusion improvements in periodontal parameters at 6 months. Although no criteria, initial sample size estimation, randomization, allocation significant difference in clinical attachment level (CAL) gain between concealment, and blinding can be found in the previously published groups was noted, the combination therapy showed a greater bone fill. paper (Saito et al., 2019). Given these results from the short-term study, we thought it important to evaluate the longevity of effects of the combination therapy. The objective of this follow-up study was to evaluate 2-year 2.3 | Clinical and radiographic examinations outcomes of the combination therapy using rhFGF-2 and DBBM in comparison with rhFGF-2 alone in the treatment of intrabony CAL, PPD, gingival recession (GR), bleeding on probing (BOP), defects. and tooth mobility (TM) were evaluated at baseline (post-initial 94 | AOKI et al. periodontal therapy; IP), 6 months (Saito et al., 2019), 1 year, and 2 years postoperatively by calibrated examiners. Standardized periapical radiographs were taken, and radiographic bone fill (RBF; %) was calculated as described previously (Seshima et al., 2017).

(a) (b) 2.4 | Patient-reported outcome measure

At each evaluation time point, participants were asked to rate the perception of oral health, using an oral health-related quality of life (QoL) instrument, OHRQL-J (Saito et al., 2010, 2011). In this study, total OHRQL-J score from the 22 items was used for analysis.

Details for surgical intervention have been provided in the previous paper (Saito et al., 2019). Briefly, full-thickness flaps were raised following infiltration anaesthesia. After debridement, scaling and root planing, and rinsing, the test sites received 0.3% rhFGF-2 (REGROTH® Dental Kit, Kaken Pharmaceutical) with DBBM (Geistlich Bio-Oss®, Geistlich Pharma AG). Defects in the control (e) (f) group received only rhFGF-2 formulation. The flaps were closed with interrupted sutures and modified vertical mattress sutures. Detailed information on postsurgical and maintenance care can be found elsewhere (Saito et al., 2019). Representative clinical cases are shown in Figure 1.

2.6 | Statistical analysis (g) (h) Clinical data from the two centres were pooled for analysis. The primary endpoint was the CAL gain at 2 years postoperatively. Fisher's exact test was employed to analyse categorical variables. Difference between two groups was sought by the Mann–Whitney U test. The Friedman test with Dunn post hoc test was used to compare intra-group data over time. Correlations between CAL gains and baseline parameters were assessed by Spearman's rank correlation. Multiple regression analysis was used to evaluate the relationship of PPD, BOP, defect depth,

(i) (j)

FIGURE 1 Clinical cases. (a–f) 60-year-old woman; received rhFGF-2 + DBBM (test group). (a) Baseline (palatal). The mesial aspect of #24 showed PPD of 7 mm. (b) Preoperative radiograph. Defect depth was 3 mm, width 5 mm (confirmed during surgery). (c) 1-year follow-up view. (d) 1-year radiograph. (e) 2-year follow-up view; PPD = 2 mm. (f) 2-year radiograph. (g–l) 53-year-old woman; received rhFGF-2 (control group). (g) Baseline. PPD at the distal aspect of #33 was 7 mm. (h) Preoperative radiograph. Defect depth was 5 mm, width 3 mm (confirmed during surgery). (i) 1-year follow- up view. (j) 1-year radiograph. (k) 2-year follow-up. PPD = 2 mm. (l) (k) (l) 2-year radiograph AOKI et al. | 95

TABLE 1 Defect locations and configurations showed a gain of 3.11 ± 1.46 mm. No significant difference was rhFGF-2 (control, rhFGF-2 + DBBM noted between groups at any time point. Intrabony defect n = 18) (test, n = 20) At 2 years, 25.0% of sites (n = 5) in the test group achieved CAL Position [n (%)] gains of >4 mm from baseline, while 33.3% (n = 6) in the control Maxilla 6 (33.3) 9 (45.0) group showed such value (Table S1). Mandible 12 (66.7) 11 (55.0) The values of PPD reductions from baseline were 3.58 ± 1.21 mm Anterior teeth 5 (27.8) 2 (10.0) in the test group and 3.58 ± 1.53 mm in the control group; no significant inter-group difference was noted. Premolars 4 (22.2) 5 (25.0) Regarding GR and TM, no significant intra- or intergroup dif- Molars 9 (50.0) 13 (65.0) ferences were found (Table 2). In both groups, the values of BOP Morphology [n (%)] positive (%) were significantly reduced at 6 months and thereafter, 1–wall 3 (16.7) 2 (10.0) when compared to baseline, while no significant difference between 2–wall 4 (22.2) 5 (25.0) groups was found at any time point. 3–wall 6 (33.3) 5 (25.0) Combination 5 (27.8) 8 (40.0) Depth (mm; 4.81 ± 1.86 (range, 4.70 ± 1.08 (range, 3.3 | Relationship between CAL gain and variables mean ± SD) 3.0–11.0) 3.0–6.5) at baseline Width (mm; 2.89 ± 0.78 (range, 3.83 ± 1.83* (range, mean ± SD) 2.0–5.0) 2.0–10.0) Next, relationships between 2-year CAL gains and baseline data Note: Mann–Whitney U test, two-tailed. were sought. Postoperative CAL gain and baseline CAL or PPD *p = 0.0403. showed a significant positive correlation in both groups (Table S2). and defect width at baseline with CAL gain from baseline to 2 years A significant positive correlation was found between CAL gain and postoperatively (dependent variable). Statistical software packages baseline defect depth in the test group. (InStat 3.10 and Prism 7.05, GraphPad Software) were used. A p The results of multiple regression analyses are shown in Tables value of 0.05 was considered statistically significant. S3 and S4. There was no multicollinearity among the variables. In the control group, the baseline PPD showed a significant relationship with the postoperative CAL gain at 2 years (Table S3). In the test 3 | RESULTS group, the defect depth at baseline showed a significant association with the postoperative CAL gain (Table S4). 3.1 | Participants and clinical parameters

In the base study (Saito et al., 2019), 32 patients participated and 3.4 | Radiographic evaluation contributed 44 sites. The progress made during the study is shown in Figure S1. At 2-year follow-up, a total of 38 sites [20 belonging In both groups, there was a progressive increase in RBF (Table 2, Figure 2b). to the test group (rhFGF-2+DBBM) and 18 to the control group At 2 years postoperatively, the mean value for RBF in the test group (rhFGF-2 alone)] in 30 patients were re-evaluated. Accidental death, (56.2%) was significantly greater compared to the control group (40.8%). serious injury, and no shows accounted for the missing scores. The participant demographics and baseline full-mouth clinical parameters can be found in the previous paper (Saito et al., 2019). 3.5 | Comparison of CAL gains and radiographic outcomes between different defect configurations

3.2 | Clinical outcomes In the control group, 3–wall defects yielded significantly greater CAL gains and RBF than 1-2 wall defects (Table 3). In contrast, no signifi- Postoperative healing occurred without significant problems. cant difference in those values between different defect configura- Characteristics of intrabony defects are shown in Table 1. Between tions was found in the test group. In 1-2-wall defects, the test group groups, there were no significant differences in maxillary and man- showed greater RBF than the control group. dibular defects, tooth type, defect configuration, or defect depth. The test group had significantly wider defects than the control group. At 1 and 2 years postoperatively, significant improvements in 3.6 | Changes in scores on OHRQL-J CAL and PPD were found in both groups (Table 2). In both treatment groups, the level of improvement found at 6 months has been No significant change in the mean total OHRQL-J scores over time sustained over a 2-year period (Figure 2a). At 2 years, the test group was found in either group (Figure 3). There was no significant inter- showed a mean CAL gain of 3.35 ± 1.28 mm, while the control group group difference at any time point. 96 | AOKI et al.

TABLE 2 Clinical and radiographic outcomes of treated sites (Total n = 38 sites)

Variable/Group Baseline (post-IP) 6 months 1 year 2 years

CAL (mm) rhFGF-2 (control) 7.19 ± 1.66 (6.5; 6.00–8.25) 4.42 ± 1.43*** (4; 4.14 ± 1.50*** (4; 2.88–5.50) 4.08 ± 1.33*** (4; 3.38–5.25) 2.88–5.00) rhFGF-2 + DBBM 7.67 ± 1.68 (7; 6.25–8.88) 4.53 ± 1.42*** (4; 4.50 ± 1.28*** (4.5; 4.28 ± 1.30*** (4; (test) 4.00–5.50) 4.00–5.00) 3.50–5.00) Diff. between groups N.S. N.S. N.S. N.S. PPD (mm) rhFGF-2 6.19 ± 1.41 (5; 5.00–7.00) 2.83 ± 0.87*** (3; 2.67 ± 0.84*** (3; 2.00–3.00) 2.61 ± 0.87*** (2; 2.00–3.00) 2.00–3.25) rhFGF-2 + DBBM 6.30 ± 1.30 (6.5; 5.00–7.00) 2.80 ± 0.73*** (3; 2.83 ± 0.54*** (3; 2.25–3.00) 2.73 ± 0.55*** (3; 2.00–3.38) 2.00–3.00) Diff. between groups N.S. N.S. N.S. N.S. GR (mm) rhFGF-2 0.94 ± 1.16 (1; 0.00–1.25) 1.36 ± 1.46 (1; 0.00–2.00) 1.47 ± 1.22 (1.5; 0.38–2.00) 1.47 ± 1.30 (1.25; 0.38–2.00) rhFGF-2 + DBBM 1.33 ± 1.42 (1; 0.00–2.75) 1.73 ± 1.36 (1.75; 1.63 ± 1.21 (1.75; 1.48 ± 1.21 (1.75; 1.00–2.00) 1.00–2.00) 0.50–2.00) Diff. between groups N.S. N.S. N.S. N.S. BOP positive (%) rhFGF-2 66.7 11.1*** 5.6*** 0.0*** rhFGF-2 + DBBM 75.0 5.0*** 0.0*** 0.0*** Diff. between N.S. N.S. N.S. N.S. groupsa TM rhFGF-2 0.11 ± 0.32 (0; 0.00–0.00) 0.06 ± 0.24 (0; 0.06 ± 0.24 (0; 0.00–0.00) 0.11 ± 0.32 (0; 0.00–0.00) 0.00–0.00) rhFGF-2 + DBBM 0.20 ± 0.41 (0; 0.00–0.00) 0.05 ± 0.22 (0; 0.00-0.00) 0.05 ± 0.22 (0; 0.00–0.00) 0.05 ± 0.22 (0; 0.00–0.00) Diff. between groups N.S. N.S. N.S. N.S. RBF (%) rhFGF-2 — 31.2 ± 13.3 (30; 36.7 ± 15.2 (34.9; 40.8 ± 17.2a (38.2; 20.6–40.4) 25.8–50.0) 27.2–54.7) rhFGF-2 + DBBM — 47.7 ± 16.8 (47.2; 54.6 ± 17.7 (62.6; 41.1–68.3) 56.2 ± 18.0† (62.6; 36.4–63.3) 45.8–69.2) Diff. between groups p = 0.004 p = 0.003 p = 0.013

Note: Data shown as mean ± standard deviation (median; interquartile range), except for BOP. Difference between groups at each time point was assessed by Mann–Whitney U test. Intra-group difference over time was assessed by Friedman test with Dunn post-test. Abbreviations: BOP, bleeding on probing; CAL, clinical attachment level; GR, gingival recession; IP, initial periodontal therapy; PPD, probing pocket depth; RBF, radiographic bone fill; TM, tooth mobility. aCategorical data were assessed by Fisher's exact test. ***p < 0.001, compared to baseline; †p < 0.05, ††p < 0.01, compared to 6 M.

test and control groups, respectively. These values were compa- 4 | DISCUSSION rable to the 2-year value of 3.3 mm from our previous study using EMD alone, with similar baseline CAL measurements (Seshima Most RCTs are relatively short term and, due to various reasons, et al., 2017). As for the CAL gain following the use of rhFGF-2, the they are seldom re-visited or extended (Davies et al., 2018). 9-month value of 2.2 mm was reported in a multicentre RCT using Obviously, there is no guarantee that treatment effects remain un- rhFGF-2 alone (Kitamura et al., 2016) and 6-month value of 3.0 mm changed beyond the initial study. In the present study, we evalu- in another RCT using rhFGF-2 with beta-tricalcium phosphate ated 2-year follow-up outcomes from our previous 6-month RCT (β-TCP; Cochran et al., 2016). The finding that no significant dif- (Saito et al., 2019). At 2 years postoperatively, the mean values ference existed between groups is in line with the controlled stud- of CAL gain, the primary endpoint, were 3.4 and 3.1 mm in the ies using EMD alone or in combination with alloplastic materials AOKI et al. | 97

(Bokan et al., 2006; Jepsen et al., 2008; Sculean et al., 2007). Our deeper pockets (Seshima et al., 2017; Tonetti et al., 2002; Zucchelli 2-year results can be interpreted that both modalities were simi- et al., 2002). These findings indicate that caution must be taken larly effective in the treatment of intrabony defects existed in the when comparing postoperative CAL values from different studies. participants. This is remarkable considering the viscous nature of Recently, Trombelli et al. (2020) proposed a novel composite the rhFGF-2 formulation, which does not particularly have space- outcome measure (COM) for periodontal regenerative therapy. making property. COM is consisted of clinically relevant CAL gain of ≥3 mm and In the present study, CAL gains were significantly associated postoperative PD ≤4 mm (pocket closure). In the present study, with baseline PPD in the control group (correlation and multiple re- both treatment groups showed a CAL gain of ≥3 mm and PD of gression analyses) and in the test group (correlation analysis). This <3 mm at 2 years, which can be regarded as “successful,” according was expected because studies have shown that generally, more to COM. One should keep in mind that, at baseline, the test group CAL gain can be expected following the regenerative treatment of had significantly wider defects (3.8 mm) than the control group (2.9 mm). Although the linear correlation and multiple regression analyses showed no significant association between CAL gain at 2 years and defect width at baseline, the difference in baseline defect width may partially account for no significant inter-group difference in CAL gain. Bone level is another important outcome measure for periodontal regenerative therapy. Cochran et al. (2016) stated that combining outcome measures for soft and hard tissues is preferable to assess the clinical performance of a biological agent with effects on both tissue types. At 2 years postoperatively, the mean value of RBF was significantly greater in the test group (56%) compared with the control group (41%). It can be argued that the test sites should show a greater RBF value, because radiopaque material was used with rhFGF-2 to fill the defect. It is, however, important to note that RBF values in the test sites also showed a progressive increase, which suggests bone formation. It has been reported that clinical results of the treatment of intrabony defects may be difficult to predict based on their characteristics (Renvert, Garrett, et al., 1985). In our analysis, the defect depth at baseline was positively correlated with the CAL gains at 2 years in the test group. In the multiple regression analysis, the defect depth could predict the level of CAL gain at 2-year follow-up. When values of CAL gain and RBF at 2-year follow-up were compared between different defect configurations at baseline, the treatment with rhFGF-2 yielded significantly greater CAL gains and RBF in 3-wall defects than 1– to 2–wall defects. This was expected FIGURE 2 Clinical attachment level (CAL) gain (a) and because 3–wall defects provide favourable environment for blood radiographic bone fill (RBF) (b). Box-and-whiskers plot showing minimum, maximum, median, and 25th and 75th percentiles. clot formation and cell migration from the remaining periodontal **p < 0.01, compared to control group; Mann–Whitney U test. tissues (Polson & Heul, 1978; Renvert, Nilvéus, et al., 1985). It has †p < 0.05, ††p < 0.01, compared to 6 M; Friedman test with Dunn been suggested that the extent and location of tissue resources, post-test cells, and vascularity surrounding the defect have an effect on the

TABLE 3 Comparison of clinical attachment level (CAL) gain and radiographic bone fill (RBF) at 2 years postoperatively between different defect configurations

Defect rhFGF−2 (control) Difference rhFGF−2 + DBBM (test) Difference

CAL gain (mm) 3–wall 3.64 ± 1.21 (3.50; 2.50 - 5.00) p = 0.037 3.50 ± 1.24 (3.25; 2.50 - 4.38) N.S. 1-2-wall 2.29 ± 1.52 (2.00; 1.00 - 3.50) 3.13 ± 1.38 (3.25; 2.00 - 4.38) RBF (%) 3–wall 52.1 ± 8.1 (50.0; 47.5 - 60.6) p = 0.034 51.3 ± 21.4 (50.0; 30.7 - 72.5) N.S. 1-2-wall 34.3 ± 18.1 (28.7; 22.7 - 48.0) 57.8 ± 17.3* (63.6; 50.0 - 66.7)

Note: Data shown as mean ± standard deviation (median; interquartile range). Difference between different defect configurations within group or difference between groups within the same defect configuration was assessed by Mann–Whitney U test (*p = 0.036, compared to the control group). 98 | AOKI et al.

with DBBM at 2-year follow-up. The combination therapy achieved a significantly greater RBF. In both treatment groups, favourable clinical and patient-centred outcomes can be sustained for at least 2 years.

ACKNOWLEDGEMENTS The authors thank Eiichi Suzuki, DDS, PhD, Takahiro Takeuchi, DDS, PhD, Asako Makino-Oi, DDS, PhD, Katsuya Noda, DDS, PhD, and Haruka Koga, DDS for clinical treatment. The authors are also grate- ful to Risa Kobayashi, RDH and the dental hygienists at Tokyo Dental College Suidobashi Hospital, for their patient care.

CONFLICT OF INTEREST There are no conflicts of interest regarding this article. FIGURE 3 Change in total OHRQL-J scores. Data shown as mean ± standard deviation. IP, initial periodontal therapy ORCID Hiroki Sugito https://orcid.org/0000-0001-9405-4197 Atsushi Saito https://orcid.org/0000-0002-0065-2207 regenerative potential (Kim et al., 2004). In case of non-contained defects, regenerative therapy in combination with bone substitutes REFERENCES is indicated (Cortellini & Tonetti, 2000). When the use of EMD with Armitage, G. C. (1999). Development of a classification system for peri- DBBM was compared to collagen barrier with DBBM in an RCT of odontal diseases and conditions. Annals of Periodontology, 4, 1–6. https://doi.org/10.1902/annals.1999.4.1.1 the treatment of deep non-contained intrabony defects, comparable Bokan, I., Bill, J. S., & Schlagenhauf, U. (2006). Primary flap closure com- clinical outcomes were noted after 12 months (Iorio-Siciliano et al., bined with Emdogain® alone or Emdogain® and Cerasorb® in the 2014). In the present study, the test treatment showed similar CAL treatment of intra-bony defects. Journal of Clinical Periodontology, gains in 1– to 2–wall and 3–wall defects at 2 years. In contrast, the 33, 885–893. https://doi.org/10.1111/j.1600-051X.2006.01010.x Camargo, P. M., Lekovic, V., Weinlaender, M., Nedic, M., Vasilic, N., control treatment yielded significantly greater CAL gains and RBF in Wolinsky, L. E., & Kenney, E. B. (2000). A controlled re-entry study 3–wall defects. Within 1-2-wall defects, the test treatment yielded on the effectiveness of bovine porous bone mineral used in combi- significantly greater RBF compared to the control. These results may nation with a collagen membrane of porcine origin in the treatment indicate that: (1) in the treatment of 3–wall defects, the sole use of of intrabony defects in humans. Journal of Clinical Periodontology, rhFGF-2 may be sufficient; (2) a greater level of healing can be ex- 27, 889–896. https://doi.org/10.1034/j.1600-051x.2000.02701 2889.x pected by adding DBBM in more challenging cases such as deeper Cochran, D. L., Oh, T.-J., Mills, M. P., Clem, D. S., McClain, P. K., Schallhorn, defects or 1– to 2–wall defects. In some studies, no additional ben- R. A., McGuire, M. K., Scheyer, E. T., Giannobile, W. V., Reddy, M. efits in clinical outcomes were found regarding the use of EMD with S., Abou-Arraj, R. V., Vassilopoulos, P. J., Genco, R. J., Geurs, N. C., bone substitutes (Hoffmann et al., 2016; Kao et al., 2015; Troiano & Takemura, A. (2016). A randomized clinical trial evaluating rh- FGF-2/ -TCP in periodontal defects. Journal of Dental Research, 95, et al., 2017). Potential benefits of adding various bone substitutes β 523–530. https://doi.org/10.1177/0022034516632497 to rhFGF-2 therapy and indications need to be verified by further Cortellini, P., & Tonetti, M. S. (2000). Focus on intrabony defects: guided studies. tissue regeneration. Periodontology 2000, 22, 104–132. https://doi. There are limitations to this study. Due to the study design and org/10.1034/j.1600-0757.2000.2220108 .x Davies, G., Jordan, S., Brooks, C. J., Thayer, D., Storey, M., Morgan, G., relatively small sample size, this study was underpowered to as- Allen, S., Garaiova, I., Plummer, S., & Gravenor, M. (2018). Long sess the effect of the number of residual bony walls in detail. Two- term extension of a randomised controlled trial of probiotics using year follow-up is still a relatively short-term observation. A longer electronic health records. Scientific Reports, 8, 7668. https://doi. observation period will be necessary. Moreover, regarding the re- org/10.1038/s41598-018-25954-z Hoffmann, T., Al-Machot, E., Meyle, J., Jervøe-Storm, P. M., & Jepsen, generative capability of rhFGF-2, further studies evaluating human S. (2016). Three-year results following regenerative periodon- histologic evidence are needed. Despite the limitations, we believe tal surgery of advanced intrabony defects with enamel ma- that this follow-up study provides relevant implications for the use trix derivative alone or combined with a synthetic bone graft. of rhFGF-2 therapy. Clinical Oral Investigations, 20, 357–364. https://doi.org/10.1007/ s00784-015-1522-4 Iorio-Siciliano, V. I., Andreuccetti, G., Blasi, A., Matarasso, M., Sculean, A., & Salvi, G. E. (2014). Clinical outcomes following regenerative 5 | CONCLUSIONS therapy of non-contained intrabony defects using a deproteinized bovine bone mineral combined with either enamel matrix derivative In the treatment of intrabony defects, no significant difference in or collagen membrane. Journal of Periodontology, 85, 1342–1350. https://doi.org/10.1902/jop.2014.130420 CAL gains was found between the use of rhFGF-2 alone and rhFGF-2 AOKI et al. | 99

Iorio-Siciliano, V. I., Andreuccetti, G., Iorio-Siciliano, A., Blasi, A., Sculean, A., & Lin, Z., Rios, H. F., & Cochran, D. L. (2015). Emerging regenera- Salvi, G. E. (2011). Clinical outcomes after treatment of non-contained tive approaches for periodontal reconstruction: A system- intrabony defects with enamel matrix derivative or guided tissue re- atic review from the AAP Regeneration Workshop. Journal of generation: A 12-month randomized controlled clinical trial. Journal of Periodontology, 86(2 Suppl), S134–S152. https://doi.org/10.1902/ Periodontology, 82, 62–71. https://doi.org/10.1902/jop.2010.100144 jop.2015.130689 Irokawa, D., Okubo, N., Nikaido, M., Shimizu, H., Konobu, N., Matsui, Matarasso, M., Iorio-Siciliano, V., Blasi, A., Ramaglia, L., Salvi, G. E., & T., Fujita, T., Goto, H., Takeuchi, T., Ishii, Y., & Saito, A. (2017). Sculean, A. (2015). Enamel matrix derivative and bone grafts for Periodontal regenerative therapy of intrabony defects using periodontal regeneration of intrabony defects. A systematic re- deproteinized bovine bone mineral in combination with collagen view and meta-analysis. Clinical Oral Investigations, 19, 1581–1593. barrier membrane: A multicenter prospective case-series study. https://doi.org/10.1007/s00784-015-1491-7 International Journal of Periodontics & Restorative Dentistry, 37, 393– Murakami, S. (2011). Periodontal tissue regeneration by signaling mole- 401. https://doi.org/10.11607/prd.2888 cule (s): What role does basic fibroblast growth factor (FGF-2) have Irokawa, D., Takeuchi, T., Noda, K., Goto, H., Egawa, M., Tomita, S., Sugito, in periodontal therapy? Periodontology 2000, 56, 188–208. https:// H., Nikaido, M., & Saito, A. (2017). Clinical outcome of periodontal doi.org/10.1111/j.1600-0757.2010.00365.x regenerative therapy using collagen membrane and deproteinized Nevins, M., Kao, R. T., McGuire, M. K., McClain, P. K., Hinrichs, J. E., bovine bone mineral: a 2.5-year follow-up study. BMC Research McAllister, B. S., Reddy, M. S., Nevins, M. L., Genco, R. J., Lynch, Notes, 10, 102. https://doi.org/10.1186/s13104-017-2426-y S. E., & Giannobile, W. V. (2013). Platelet-derived growth factor Jepsen, S., Topoll, H., Rengers, H., Heinz, B., Teich, M., Hoffmann, T., promotes periodontal regeneration in localized osseous defects: Al-Machot, E., Meyle, J., & Jervøe-Storm, P.-M. (2008). Clinical 36-month extension results from a randomized, controlled, dou- outcomes after treatment of intra-bony defects with an EMD/ ble-masked clinical trial. Journal of Periodontology, 84, 456–464. synthetic bone graft or EMD alone: A multicentre randomized-con- https://doi.org/10.1902/jop.2012.120141 trolled clinical trial. Journal of Clinical Periodontology, 35, 420–428. Page, R. C., & Eke, P. I. (2007). Case definitions for use in popula- https://doi.org/10.1111/j.1600-051X.2008.01217.x tion-based surveillance of periodontitis. Journal of Periodontology, Kao, R. T., Nares, S., & Reynolds, M. A. (2015). Periodontal regeneration – 78, 1387–1399. https://doi.org/10.1902/jop.2007.060264 Intrabony defects: a systematic review from the AAP Regeneration Polson, A. M., & Heul, L. C. (1978). Osseous repair in infrabony periodon- Workshop. Journal of Periodontology, 86(2 Suppl), S77–S104. tal defects. Journal of Clinical Periodontology, 5, 13–23. https://doi. https://doi.org/10.1902/jop.2015.130685 org/10.1111/j.1600-051X.1978.tb01902.x Khoshkam, V., Chan, H. L., Lin, G. H., Mailoa, J., Giannobile, W. V., Wang, Renvert, S., Garrett, S., Nilvéus, R., Chamberlain, A. D. H., & Egelberg, H. L., & Oh, T. J. (2015). Outcomes of regenerative treatment with J. (1985). Healing after treatment of periodontal intraosseous rh PDGF-BB and rh FGF-2 for periodontal intra-bony defects: a sys- defects: VI. Factors influencing the healing response. Journal of tematic review and meta-analysis. Journal of Clinical Periodontology, Clinical Periodontology, 12, 707–715. https://doi.org/10.1111/ 42, 272–280. https://doi.org/10.1111/jcpe.12354 j.1600-051X.1985.tb01396.x Kim, C. S., Choi, S. H., Chai, J. K., Cho, K. S., Moon, I. S., Wikesjö, U. Renvert, S., Nilvéus, R., & Egelberg, J. (1985). Healing after treatment of M., & Kim, C. K. (2004). Periodontal repair in surgically created in- periodontal intraosseous defects: V. Effect of root planing versus trabony defects in dogs: influence of the number of bone walls on flap surgery. Journal of Clinical Periodontology, 12, 619–629. https:// healing response. Journal of Periodontology, 75, 229–235. https:// doi.org/10.1111/j.1600-051X.1985.tb00933.x doi.org/10.1902/jop.2004.75.2.229 Reynolds, M. A., Kao, R. T., Nares, S., Camargo, P. M., Caton, J. G., Clem, Kitamura, M., Akamatsu, M., Kawanami, M., Furuichi, Y., Fujii, T., Mori, M., D. S., Fiorellini, J. P., Geisinger, M. L., Mills, M. P., Nevins, M. L., Kunimatsu, K., Shimauchi, H., Ogata, Y., Yamamoto, M., Nakagawa, & Rosen, P. S. (2015). Periodontal regeneration – Intrabony de- T., Sato, S., Ito, K., Ogasawara, T., Izumi, Y., Gomi, K., Yamazaki, K., fects: practical applications from the AAP regeneration workshop. Yoshie, H., Fukuda, M., … Murakami, S. (2016). Randomized place- Clinical Advances in Periodontics, 5, 21–29. https://doi.org/10.1902/ bo-controlled and controlled non-inferiority Phase III trials compar- cap.2015.140062 ing trafermin, a recombinant human fibroblast growth factor 2, and Saito, A., Bizenjima, T., Takeuchi, T., Suzuki, E., Sato, M., Yoshikawa, enamel matrix derivative in periodontal regeneration in intrabony K., Kitamura, Y., Matsugami, D., Aoki, H., Kita, D., Imamura, K., defects. Journal of Bone and Mineral Research, 31, 806–814. https:// Irokawa, D., Seshima, F. & Tomita, S. (2019). Treatment of intrabony doi.org/10.1002/jbmr.2738 periodontal defects using rhFGF-2 in combination with deprotein- Kitamura, M., Akamatsu, M., Machigashira, M., Hara, Y., Sakagami, R., ized bovine bone mineral or rh FGF-2 alone: A 6-month random- Hirofuji, T., Hamachi, T., Maeda, K., Yokota, M., Kido, J., Nagata, ized controlled trial. Journal of Clinical Periodontology, 46, 332–341. T., Kurihara, H., Takashiba, S., Sibutani, T., Fukuda, M., Noguchi, T., https://doi.org/10.1111/jcpe.13086 Yamazaki, K., Yoshie, H., Ioroi, K., … Murakami, S. (2011). FGF-2 Saito, A., Hosaka, Y., Kikuchi, M., Akamatsu, M., Fukaya, C., Matsumoto, stimulates periodontal regeneration: results of a multi-center ran- S., Ueshima, F., Hayakawa, H., Fujinami, K., & Nakagawa, T. (2010). domized clinical trial. Journal of Dental Research, 90, 35–40. https:// Effect of initial periodontal therapy on oral health–related quality of doi.org/10.1177/00220 3 4 5 1 0 3 8 4 6 1 6 life in patients with periodontitis in Japan. Journal of Periodontology, Kitamura, M., Nakashima, K., Kowashi, Y., Fujii, T., Shimauchi, H., Sasano, 81, 1001–1009. https://doi.org/10.1902/jop.2010.090663 T., Furuuchi, T., Fukuda, M., Noguchi, T., Shibutani, T., Iwayama, Saito, A., Ota, K., Hosaka, Y., Akamatsu, M., Hayakawa, H., Fukaya, C., Y., Takashiba, S., Kurihara, H., Ninomiya, M., Kido, J., Nagata, T., Ida, A., Fujinami, K., Sugito, H., & Nakagawa, T. (2011). Potential Hamachi, T., Maeda, K., Hara, Y., … Murakami, S. (2008). Periodontal impact of surgical periodontal therapy on oral health-re- tissue regeneration using fibroblast growth factor-2: Randomized lated quality of life in patients with periodontitis: a pilot study. controlled phase II clinical trial. PLoS One, 3, e2611. https://doi. Journal of Clinical Periodontology, 38, 1115-1121. https://doi. org/10.1371/journal.pone.0002611 org/10.1111/j.1600-051X.2011.01796.x Li, F., Yu, F., Xu, X., Li, C., Huang, D., Zhou, X., Ye, L., & Zheng, L. Sculean, A., Nikolidakis, D., Nikou, G., Ivanovic, A., Chapple, I. L. C., & (2017). Evaluation of recombinant human FGF-2 and PDGF-BB Stavropoulos, A. (2015). Biomaterials for promoting periodon- in periodontal regeneration: A systematic review and me- tal regeneration in human intrabony defects: a systematic re- ta-analysis. Scientific Reports, 7, 1–10. https://doi.org/10.1038/ view. Periodontology 2000, 68, 182–216. https://doi.org/10.1111/ s41598-017-00113-y prd.12086 100 | AOKI et al.

Sculean, A., Nikolidakis, D., & Schwarz, F. (2008). Regeneration of peri- review, meta-analysis and trial sequential analysis. Journal of Clinical odontal tissues: combinations of barrier membranes and grafting Periodontology, 44, 729–738. https://doi.org/10.1111/jcpe.12742 materials–biological foundation and preclinical evidence: a system- Trombelli, L., Farina, R., Vecchiatini, R., Maietti, E., & Simonelli, A. (2020). atic review. Journal of Clinical Periodontology, 35, 106–116. https:// A simplified composite outcome measure to assess the effect doi.org/10.1111/j.1600-051X.2008.01263.x of periodontal regenerative treatment in intraosseous defects. Sculean, A., Pietruska, M., Arweiler, N. B., Auschill, T. M., & Nemcovsky, Journal of Periodontology, 91, 723–731. https://doi.org/10.1002/ C. (2007). Four-year results of a prospective controlled clinical JPER.19-0127 study evaluating healing of intrabony defects following treatment Zucchelli, G., Amore, C., Montebugnoli, L., & De Sanctis, M. (2003). with an enamel matrix protein derivative alone or combined with Enamel matrix proteins and bovine porous bone mineral in the a bioactive glass. Journal of Clinical Periodontology, 34, 507–513. treatment of intrabony defects: a comparative controlled clin- https://doi.org/10.1111/j.1600-051X.2007.01084.x ical trial. Journal of Periodontology, 74, 1725–1735. https://doi. Seshima, F., Aoki, H., Takeuchi, T., Suzuki, E., Irokawa, D., Makino-Oi, A., org/10.1902/jop.2003.74.12.1725 Sugito, H., Tomita, S., & Saito, A. (2017). Periodontal regenerative Zucchelli, G., Bernardi, F., Montebugnoli, L., & De Sanctis, M. (2002). therapy with enamel matrix derivative in the treatment of intra- Enamel matrix proteins and guided tissue regeneration with tita- bony defects: a prospective 2-year study. BMC Research Notes, 10, nium-reinforced expanded polytetrafluoroethylene membranes 256. https://doi.org/10.1186/s13104-017-2572-2 in the treatment of infrabony defects: a comparative controlled Stavropoulos, A., & Karring, T. (2010). Guided tissue regeneration com- clinical trial. Journal of Periodontology, 73, 3–12. https://doi. bined with a deproteinized bovine bone mineral (Bio-Oss®) in the org/10.1902/jop.2002.73.1.3 treatment of intrabony periodontal defects: 6-year results from a randomized-controlled clinical trial. Journal of Clinical Periodontology, 37, 200–210. https://doi.org/10.1111/j.1600-051X.2009.01520.x SUPPORTING INFORMATION Tonetti, M. S., Cortellini, P., Lang, N. P., Suvan, J. E., Adriaens, P., Dubravec, Additional supporting information may be found online in the D., & Zybutz, M. (2004). Clinical outcomes following treatment of Supporting Information section. human intrabony defects with GTR/bone replacement material or access flap alone. Journal of Clinical Periodontology, 31, 770–776. https://doi.org/10.1111/j.1600-051X.2004.00562.x Tonetti, M. S., Lang, N. P., Cortellini, P., Suvan, J. E., Adriaens, P., Dubravec, D., Fonzar, A., Fourmousis, I., Mayfield, L., Rossi, R., How to cite this article: Aoki H, Bizenjima T, Seshima F, et al. Silvestri, M., Tiedemann, C., Topoll, H., Vangsted, T., & Wallkamm, Periodontal surgery using rhFGF-2 with deproteinized bovine B. (2002). Enamel matrix proteins in the regenerative therapy of bone mineral or rhFGF-2 alone: 2-year follow-up of a deep intrabony defects: a multicenter randomized controlled clinical trial. Journal of Clinical Periodontology, 29, 317–325. https://doi. randomized controlled trial. J Clin Periodontol. 2021;48:92– org/10.1034/j.1600-051X.2002.290407.x 100. https://doi.org/10.1111/jcpe.13385 Troiano, G., Laino, L., Zhurakivska, K., Cicciù, M., Lo Muzio, L., & Lo Russo, L. (2017). Addition of enamel matrix derivatives to bone substitutes for the treatment of intrabony defects: A systematic Received: 19 June 2020 | Revised: 5 September 2020 | Accepted: 29 September 2020 DOI: 10.1111/jcpe.13381

SYSTEMATIC REVIEW

Periodontal infrabony defects: Systematic review of healing by defect morphology following regenerative surgery

1Periodontology Unit, Centre for Host- Microbiome Interactions, Faculty of Abstract Dentistry, Oral & Craniofacial Sciences, Background: It is thought that infrabony defect morphology affects the outcome King’s College London, London, UK 2Universita’ di Foggia, Foggia, Italy of periodontal regenerative surgery. However, this has not been systematically 3Division of Periodontology and Implant investigated. Dentistry, Faculty of Dentistry, The Aims: To investigate how well defect morphology is described in papers reporting University of Hong Kong, Hong Kong, Hong Kong regenerative therapy of periodontal infrabony defects and to investigate its effect on 4University of California, San Francisco, clinical and radiographic outcomes. USA Materials and Methods: A search was conducted in 3 electronic databases for pub- 5Department of Oral and Maxillofacial Implantology, Shanghai Key Laboratory of lications reporting clinical and radiographic outcomes of periodontal intra-bony de- Stomatology, National Clinical Research fects after regenerative therapy, divided by defect morphology. Centre of Stomatology, Shanghai 9th People Hospital, School of Medicine Results: The initial search resulted in 4487 papers, reduced to 143 after first and sec- Shanghai Jiao Tong University, Shanghai, ond screening. Fifteen of these publications were suitable for a fixed-effects meta- China analysis. Initial defect depth was found to influence radiographic bone gain 12 months Correspondence post-surgery, while narrower angles and increased number of walls influenced both Luigi Nibali, King's College London Dental Institute, Guy's Hospital, Great Maze radiographic bone gain and clinical attachment level (CAL) gain at 12 months. These Pond, London, UK. associations seemed to occur irrespective of biomaterials used. Risk of bias ranged Email: [email protected] from low to high. Conclusion: Deeper defects with narrower angles and increased number of walls exhibit improved CAL and radiographic bone gain at 12 months post-regenerative surgery. More data are needed about other aspects of defect morphology such as extension to buccal/lingual surfaces.

KEYWORDS intra-bony, osseous defect, periodontitis, regeneration

1 | INTRODUCTION “infraosseous” (“infrabony”) (Goldman & Cohen, 1958). These authors defined suprabony defects as those where the base of the Classically, periodontal defects have been differentiated based on pocket is located coronal to the alveolar crest. On the other hand, bone resorption patterns into “supraosseous” (“suprabony”) and infrabony defects are those with apical location of the base of the

wileyonlinelibrary.com/journal/jcpe | 101 J Clin Periodontol. 2021;48:101–114. 102 | NIBALI et al. pocket relative to the bone crest. Goldman and Cohen then classified infrabony defects according to the location and number of os- Clinical relevance seous walls remaining around the pocket. It has been suggested that Scientific rationale for the study: It is important to establish the term “intrabony” means “within or inside the bone”, while “infra- which aspects of infrabony defect morphology influence bony” means “below the crest of bone” (Weinberg & Eskow, 2000). outcomes of regenerative surgery. These authors suggested that only 3-wall angular defects should be Principle findings: Defect depth, angle and number of walls termed “intrabony”, while all other vertical bony defects should be appear to influence the healing following regenerative per- referred to as “infrabony”. A large body of clinical and histological iodontal surgery, irrespective of biomaterials used. evidence accumulated over the last 4-5 decades shows how heal- Practical implications: It is important to consider defect ing following periodontal surgery of infrabony defects can, with the morphology when planning regenerative periodontal use of biomaterials, be guided towards the formation of de novo surgery. More data about several aspects of defect mor- cementum, functionally oriented periodontal ligament, new alveo- phology should be routinely collected and correlated with lar bone and gingiva (Melcher, 1976; Nyman et al., 1982; Wikesjo & treatment response, in order to improve the clinician's abil- Nilveus, 1990). The emphasis on “3-wall” bony defects was due to ity to maximize healing. their higher chances of successful regeneration (Weinberg & Eskow, 2000). Recent developments in periodontal regenerative techniques and materials have pushed the boundaries of what is considered “re- generable” (P. Cortellini et al., 2020). Papapanou & Tonetti differ- following periodontal surgery has not been investigated systemati- entiated osseous defects into “suprabony” defects, “infrabony” and cally, perhaps owing to the lack of a clear classification system for “interradicular or furcation” defects (Papapanou & Tonetti, 2000). osseous defects. Therefore, the aim of this review was to examine Infrabony defects were further divided into “intrabony” and “cra- the relationship between intra-bony defect morphology and treat- ters”, and the former was subdivided into “1-, 2- or 3-wall defects” or ment outcomes. “combinations”. The emphasis was placed on differentiating whether or not the defect affected more or less to the same extent the adjacent root surfaces of an inter-dental space (similar periodontal 2 | MATERIALS AND METHODS breakdown along the root surface of two adjacent teeth, i.e. craters) or primarily affected one of the two root surfaces of an inter-dental A systematic review protocol was written in the planning stages, and space (greater periodontal breakdown on the tooth with the defect the PRISMA statement (Moher et al., 2009) was followed both in and more coronal crest of bone on the adjacent tooth in the same the planning and in the reporting of the review (checklist attached inter-dental space, i.e. intra-bony defect) (Papapanou & Tonetti, as Data S1). The protocol was registered on 26/03/2020 with 2000). PROSPERO (available from ID: 176697). Infrabony defects have been associated with risk of periodontal progression in the absence of the appropriate therapy, but not if included in regular maintenance care programmes (Heins et al., 1989; 2.1 | Focused questions Papapanou & Wennstrom, 1991)(Pontoriero et al., 1988). With the currently available regeneration procedures, materials The present review aimed to answer two focused questions: and technologies, intra-bony defects can be successfully regener- ated, subject to patient factors such as plaque control, smoking and • How often and how well is defect morphology described in pa- medical history, as well as tooth mobility, restorative and endodon- pers reporting regenerative therapy of periodontal infrabony de- tic condition (Nibali et al., 2019). Several publications reported on fects (defect depth, number of walls, extension of the defect and the superiority of periodontal regenerative therapy in the treatment defect angle)? of intra-bony defects over the conventional surgeries such as peri- • How does defect morphology predict the outcomes of regenera- odontal access flap, known as open-flap debridement surgery (OFD), tive therapy of periodontal infrabony defects? in terms of probing pocket depth (PPD) and clinical attachment loss (CAL) reductions (Castro et al., 2017; Needleman et al., 2006; Nibali et al., 2019). 2.2 | Eligibility criteria A few studies have also investigated the healing potential of infrabony defects following periodontal regeneration in relation to In brief, the PECOS method was the following: defect architecture, suggesting that narrower defects surrounded by higher numbers of bony walls have higher regenerative poten- • (P) Participants: Adult human patients with periodontitis who tial (Ellegaard & Loe, 1971; Selvig et al., 1993; Tsitoura et al., 2004). have completed a cycle of non-surgical periodontal therapy and However, the effect of defect morphology on treatment outcomes present with residual pockets and infrabony defects. NIBALI et al. | 103

• (E) Exposure: Defect morphology (depth, angle and number of summarized in order to draw a conclusion of the overall risk of bias walls) in defects undergoing mucoperiosteal surgery including re- within and across trials. This judgment was made independently generative surgery with guided tissue regeneration (GTR), enamel by two reviewers (DS, CA); any discrepancies were resolved by matrix derivative (EMD), bone fillers or substitutes, growth fac- discussion. tors (GF) or combination. • (C) Comparisons: Different types of intra-bony defect morphology and different types of biomaterials used. 2.6 | Summary measures and planned • (O) Outcomes: CAL gain, PPD reduction and radiographic bone method of analysis gain. • (S) Studies: Randomized controlled trials (RCTs), cohort studies or Studies were initially divided by reporting defect morphology (based case series. on criteria above). Among publications reporting defect morphology (even if just one aspect of defect morphology was reported), The following additional criteria were considered: study outcomes were investigated and compared with type of re- Inclusion criteria: i) definition of periodontal infrabony defects generative materials used (when possible) and by defect morphology at least 3 mm deep; ii) with at least 12-month follow-up; and iii) only reported. A meta-analysis was considered appropriate and was per- studies published in English. Exclusion criteria: i) animal studies; ii) formed in the presence of at least two studies with the same follow- reviews; iii) including less than 20 patients; and iv) studies including up and reporting the same data. The outcomes of interest were CAL patients with diabetes or immunocompromised. gain, PPD reduction and radiographic bone gain. The impact of initial defect depth, defect walls and defect angle on bone gain and CAL gain were pooled, and weighted mean difference (WMD) was esti- 2.3 | Information sources and Search mated using a computer program [Review Manager (RevMan). version 5.0. Copenhagen; The Nordic Cochrane Centre, The Cochrane Papers were searched on MEDLINE, Cochrane and Scopus data- Collaboration, 2008]. In addition, the coefficient estimates and the bases (search details are reported in Data S2). standard errors of the investigated variables, including defect depth, defect angle and number of walls, from each publication were also pooled to assess the odds ratio (OR) and 95% confidence interval (CI) 2.4 | Study characteristics of the primary and secondary outcomes. The contribution of each article was weighed. Forest plots were produced to graphically rep- This systematic review focused specifically on intra-bony defect resent the difference in outcomes. A p-value = 0.05 was used as the morphology and on its impact on regenerative treatment outcomes. level of significance. Heterogeneity was assessed with p-value for Data extraction was performed in duplicate (authors DS and CA) in- chi-square test. Random-effects meta-analyses of the selected stud- cluding description of the infrabony defect and treatment outcomes ies were applied if the p-value for chi-square test was >0.05. Fixed- by defect depth, defect angle and defect type (1-wall, 2-wall, 3-wall effects meta-analyses were applied if the p-value for chi-square or more description if available). test was ≤0.05 to avoid any bias being caused by methodological The exposure of “defect morphology” was assessed as follows: differences among studies. In addition, the funnel plot was used to assess the presence of the publication bias. The reporting of these • Description of defect depth and width/angle. meta-analyses adhered to the PRISMA (Preferred Reporting Items • Description of number of defect walls, divided into craters, 1-wall, for Systematic Review and Meta-Analyses) statement (Liberati et al., 2-wall, 3-wall or combination. 2009). • Description of extension of defect to buccal and/or lingual walls, for example following the definition of trench (Karn et al., 1984) or circumferential defects or “moats” (Karn et al., 1984). 2.7 | Evaluation of the strength of evidence • Description of materials used. • Description of study outcomes (clinical, radiographic, patient-re- Evidence provided by RTCs was rated using different levels of ported) divided by defect type and materials used. methodological strength modified from GRADE (Grading of Recommendations Assessments, Development and Evaluation) (Guyatt et al., 2008). Three different strength of evidence were 2.5 | Risk-of-bias analysis considered:

In order to assess the quality of the included studies, risk of bias • High: At least 3 RCTs at low risk of bias and low heterogeneity (I2 was assessed using the Cochrane Collaborations Tool for RCTs, < 30%). the Newcastle Ottawa Tool for cohort studies and the Modified • Moderate: More than 1 RCT and at least 1 RCT at low risk of bias Delphi Tool for case series. Assessment across all key domain was and low heterogeneity. 104 | NIBALI et al.

• Low: L ack of RC Ts or RC Ts at high risk of bias or high heterogeneit y. Defect angle: 36 papers reported data on defect width/angle. Ten papers reported treatment outcomes by defect angle (see Table 2). However, these studies could not be meta-analysed together, owing 3 | RESULTS to heterogeneity in reporting data. Figure 2 reports forest plots for meta-analysis of defect angle data. Two studies (n = 91 sites) were Data S3 presents the flow chart from initial search to included pa- included in meta-analysis of the regression estimates for the effect pers. The initial search generated 4487 articles from MEDLINE, of initial defect angle on radiographic bone gain at 12 months. A Cochrane Library and Scopus combined. After screening the titles statistically significant association was found between defect angle and abstracts, 404 articles were considered potentially suitable by <37° and increased bone gain (0.94 mm, 95% CI 0.48, 1.39) with at least one reviewer and qualified for full-text screening. Following moderate heterogeneity (I2 = 50%) (Figure 2A, categorical analysis full-text reading, 143 articles (from 136 original publications) met with 37° threshold). Three studies (n = 201 sites) were included in the defined inclusion criteria, while 261 were excluded. The reasons meta-analysis of the regression estimates for the effect of initial for exclusion are detailed in Data S3. A total of 117 RCTs, 20 cohort defect angle on radiographic bone gain at 12 months, showing no studies and 6 case series were included. The publication year ranged statistically significant associations (Figure 2B, continuous analysis). from 1992 to 2019. Cohen's kappa value for inter-reviewer agree- Four studies (n = 274 sites) were included in meta-analysis of the ment was 0.57 at title and abstract screening level (94.7% agreement) regression estimates for the effect of initial defect angle on CAL gain and 0.87 at second screening (94.3% agreement). Every effort was at 12 months, showing a statistically significant association between made to retrieve original data from authors when needed. Data from narrower angles and increased CAL gain (OR = 0.97, 95% CI = 0.95, 15 publications were available for meta-analyses. One study data 0.98) with low–moderate heterogeneity (I2 = 33%) (Figure 2C). (Cosyn et al., 2012) were retrieved from the study team but excluded Number of walls: 122 papers reported defect morphology as de- from the meta-analysis due to high heterogeneity detected through scribed below (see Data S5): subgroup analysis (CAL gain and bone gain data had opposite direc- tions of association based on defect morphology). Therefore, only • 1-, 2- and 3-walled defects or combinations: 78 papers 14 publications were included for quantitative analyses as detailed • 1- or 2-walled defects or combinations only: 14 papers below for the different analyses. • 1-, 2-walled, combined 1-2 or circumferential defects: 5 papers • 2-walled defects only: 1 paper • 2- or 3-walled defects or combinations: 19 papers 3.1 | Effect of defect morphology on • 3-walled defects only: 1 paper regenerative outcomes • 1-walled defects only (or mainly 1-walled): 4 papers

Defect depth: 114 papers reported average or range of defect depth Out of 122 papers reporting defect morphology, 87 reported measured radiographically and/or intra-surgically (often subdivided breakdowns of different types of defects included (based on de- by study arm). Sixteen papers (from 15 publications) presented sepa- fect morphology details above), while 35 did not. Out of 87 papers rate results for defects of different depth (see Table 1). Figure 1 reporting breakdowns of different defects by number of walls, reports forest plots for meta-analysis of defect depth data. Meta- only 17 reported treatment outcomes for defects divided by analysis was carried out for the effect of initial defect depth >4 mm baseline defect morphology. However, in 3 cases (Briguglio et al., vs. ≤4 mm on radiographic bone gain (in mm) at 12 months, including 3 2013; Crea et al., 2008; Xu et al., 2019), only 1 type of defect was papers (125 sites). A statistically significant association was found be- included in the study, so no comparison across different types of tween defect depth >4 mm and increased bone gain (−0.75 mm, 95% defects was possible. The remaining 14 papers are reported in CI −1.12, −0.38) with moderate heterogeneity (I2 = 57%) (Figure 1 A, Table 3. categorical analysis with 4 mm threshold). Six publications (includ- Figure 3 reports forest plots for meta-analysis of number of ing 314 sites) were included in meta-analysis of the regression es- walls data. Three publications (n = 150 sites) reported data on bone timates for the effect of initial defect depth on radiographic bone gain after regenerative treatment between 1-wall and 2-wall de- gain at 12 months, showing a statistically significant association be- fects and showed significant radiographic bone gain at 12 months tween deeper defect depth and increased bone gain (OR = 1.32, 95% favouring 2-wall defects (−0.57 mm, 95% CI = −0.93, −0.21) with CI = 1.19, 1.47, I2 = 0) (Figure 1 B, continuous analysis). Subgroup anal- low heterogeneity (I2 = 0%) (Figure 3 A). Two publications (n = 108 ysis by studies using GTR or EMD (with or without adjuncts) showed sites) reported data on radiographic bone gain 12 months after a significant association between deeper defect depth and increased regenerative treatment between 2-wall and 3-wall defects and bone gain of similar magnitude for both (see Data S4). Four studies showed significant bone gain favouring 3-wall defects (−0.39 mm, (n = 292 sites) were included in meta-analysis of the regression esti- 95% CI = −0.78, −0.01) with moderate heterogeneity (I2 = 54%) mates for the effect of initial depth on CAL gain at 12 months, show- (Figure 3 B). In addition, only one publication (Cortellini et al., ing no statistically significant associations (Figure 1 C). 1993) reported data on radiographic bone gain 12 months after NIBALI et al. | 105

TABLE 1 Details of papers reporting treatment outcomes divided by baseline defect depth

Author Study characteristics Results by intra-bony defect depth

Sanz et al., 2004 EMD vs. GTR (Guidor) Intra-bony defect depth did not influence significantly CAL gain. Estimate = −0.4 ± 0.2, p-value = 0.07 Meyle et al., 2011 EMD/synthetic bone Deeper intra-bony defect depth was associated with more defect fill graft vs. EMD (estimate = 3.068, p-value = 0.003) Loos et al., 2002 GTR (Guidor)/ Intra-bony defect depth did not significantly influence bone gain (p-value = 0.38) antibiotic vs. GTR alone vs. OFD/ antibiotic vs. OFD alone Grusovin & Esposito, 2009 EMD vs. placebo Initial intra-bony defect depth did not significantly influence CAL gain and radiographic bone gain at 1 year. p-value = 0.41 and 0.81, respectively Ehmke et al., 2003 GTR (Guidor) Intra-bony defect depth significantly influenced alveolar bone gain (b-weight ±SD = 0.32 ± 0.15, p-value = 0.045) Tonetti et al., 1996 GTR Titanium ePTFE Borderline significance for initial intra-bony defect depth on CAL gain at 1 year vs. GTR ePTFE vs. (p-value = 0.055) OFD Klein et al., 2001 GTR (ePTFE/ Statistically significant positive influence of baseline intra-bony depth on bone gain bioabsorbable) (p-value = <0.0001). More bone fill for initially deep intra-bony defects (≥3 mm) but no association with CAL gain Eickholz et al., 2004 GTR (ePTFE/ Deep (≥4 mm) infrabony defects exhibited statistically significantly more bioabsorbable) favourable bony fill than defects <4 mm (bone fill: 2.50 ± 1.99 and −0.57 ± 2.16, respectively). Intra-bony defect depth had statistically significant positive influence on bone fill (estimate = 0.314, p-value = 0.033) Eickholz et al., 2014 EMD/doxycycline vs. Baseline intra-bony defect depth influenced bone gain positively (p-value = 0.04) EMD/placebo Francetti et al., 2005 EMD vs. OFD Statistically significant difference in bone gain between the EMD and OFD groups only at 12 months in the ≤6-mm subgroup (p-value = 0.05, in favour to EMD) Bratthall et al., 2001 Ready-to-use EMD gel Higher CAL gain in sites with deeper baseline defects. Defects gaining >4 mm had vs. marketed EMD deeper bony defects at baseline compared to the other group which gained ≤4 mm (baseline defect depth: 6.1 ± 1.27 and 5.3 ± 1.33, respectively) Tonetti et al., 2004 GTR (Bio-Guide)/Bio- Depth of the intra-bony component did not have a significant impact on CAL gain Oss vs. OFD (estimate = 0.01 ± 0.09, p-value = 0.8751) Zucchelli et al., 2003 EMP/bovine porous Intra-bony defect depth significantly influenced CAL gain (more CAL gain in cases bone mineral vs. with deeper intra-bony component, F-ratio = 19.62, p-value = 0.00001) EMP Zucchelli et al., 2002 EMD vs. GTR (ePTFE) Intra-bony defect depth did not influence significantly CAL gain. F-ratio = 2.01, vs. OFD p-value = 0.1603 Linares et al. 2006 (same GTR (collagen Initial radiographic intra-bony defect depth was a significant covariate (p- clinical study as Tonetti membrane)/ value = 0.0001) to predict bone gain after 1 year et al., 2004) deproteinized bovine bone mineral vs. OFD Ilgenli et al., 2007 Demineralized Initial defect depth was positively correlated to the bony fill (p-value = 0.047 freeze-dried bone allograft/PRP vs. PRP

regenerative treatment between 1-wall and 3-wall defects and re- estimates for CAL gain at 12 months including five papers (n = 431 ported a significant bone gain favouring 3-wall defects (−1.18 mm, sites) revealed a statistically significant association between more 95% CI = −1.66, −0.71). bone walls and increased CAL gain (OR = 1.42, 95% CI = 1.14, 1.77) Meta-analysis of regression estimates for radiographic bone gain with high heterogeneity (I2 = 80%) (Figure 3D). Results of meta-anal- at 12 months including two studies (n = 101 sites) revealed a statis- yses including Cosyn et al. (2012) are reported in Data S6. tically significant association (OR = 3.43, 95% CI = 1.09, 10.85) with Not enough publications were available to analyse the outcome low heterogeneity (I2 = 0%) (Figure 3C). Meta-analysis of regression PPD. 106 | NIBALI et al.

(A)

(B)

(C)

FIGURE 1 Forest plots of meta-analysis of effect of defect depth on healing following regenerative surgery: categorical analysis for the effect of defect depth >4 mm on radiographic bone gain at 12 months (1A), regression estimates for the effect of initial defect depth on radiographic bone gain at 12 months (1B) and regression estimates for the effect of initial depth on CAL gain at 12 months (1C)

Overall summary of results for defect depth, angle and number 4 | DISCUSSION of walls, with associated level of evidence, is reported in Table 4. This systematic review investigated the effect of infrabony defect morphology on outcomes of periodontal regenerative therapy. The 3.2 | Risk-of-bias analysis effect of baseline defect depth, defect angle and number of walls on radiographic bone and CAL gain was investigated. This review Data S7, 8 and 9 report risk-of-bias analysis for RCT, cohort studies produced probably the largest body of systematically assessed evi- and case series. Risk-of-bias score for RCTs ranged from 0 to 10. dence to suggest that deeper, narrower defects and defects with A total of 75 papers were defined as “good quality”, 27 as “fair” more walls are associated with improved clinical and radiographic quality and 15 as “poor” quality. Aspects which recorded highest outcomes 12 months post-regenerative surgery. risk-of-bias scores were allocation concealment bias and perfor- High strength of evidence suggests that deeper defects are mance bias, while the area that according to our scoring showed associated with more radiographic bone gain at 12 months (both lowest score was reporting bias. Risk-of-bias scores for cohort continuous and categorical analyses). The magnitude of additional studies ranged from 5 to 8 stars, with the item “comparability” radiographic bone gain was approximately 0.7 mm for defects ini- often scored as 0. Eleven papers were identified as “good quality” tially deeper than 4 mm compared with those 3-4 mm deep. It is and the remaining 9 as “poor quality”. Risk-of-bias scores for case interesting to notice that the association between defect depth and series ranged from 11 to 16 out of 18. The item “study population” bone gain seems to occur irrespective of biomaterials used and was was often scored low. of the same magnitude for EMD (including studies with or without NIBALI et al. | 107

TABLE 2 Details of papers reporting treatment outcomes divided by baseline defect width/angle

Author Study characteristics Results by intra-bony defect width/angle

Losada et al., 2017 EMD/biphasic calcium phosphate vs. EMD Probability of gaining ≥3 mm of attachment diminished as the angulation score increased (OR = 2.57 higher if the treatment was performed in an angle <24.75 than in wider angles, but not statistically significant) Cortellini et al., 2008 EMD CAL gain significantly associated with the baseline radiographic defect angle (estimate = −0.05, SE = 0.02, p-value = 0.0038) Loos et al., 2002 GTR (Guidor)/antibiotic vs. GTR alone vs. Radiographic defect angle did not influence significantly bone gain OFD/antibiotic vs. OFD alone (p-value = 0.20) Ilgenli et al., 2007 Demineralized freeze-dried bone No significant differences between narrow and wide defects in allograft/PRP vs. PRP the PRP-alone therapy group (p-value = 0.89 for CAL gain & 0.90 for defect bone fill). More CAL gain and defect bone fill in favour of the narrow defects (p-value was 0.03 for both) in the DFDBA +PRP group Eickholz et al., 2004 GTR (ePTFE/bioabsorbable) Initially narrow (angle <37°) defects exhibited statistically significantly more favourable bony fill than did wide defects (bone fill: 2.30 ± 1.88 mm and −0.72 ± 2.49 mm, respectively). Baseline defect angle had statistically significant positive influence on bone fill. Estimate = −0.064, p-value = 0.003 Zucchelli et al., 2003 EMP/bovine porous bone mineral vs. EMP Defect angle did not influence significantly CAL gain (F-ratio = 2.20, p-value = 0.1439) Tonetti et al., 1993 GTR (Gore-Tex) Defect angle significantly affected CAL gain and bone gain (estimate = −0.05/p-value = 0.0026, and estimate = −0.05/p- value = 0.0031, respectively) Tsitoura et al., 2004 EMD vs. OFD Radiographic defect angle statistically significantly associated with CAL gain (p-value = 0.0477). The probability of obtaining CAL gain >3 mm was 2.464 times higher (with a 95% confidence interval: 1.017- 5.970) when the radiographic defect angle was ≤22° than when the radiographic defect angle was ≥36° Klein et al., 2001 GTR (ePTFE/bioabsorbable) Initially, narrow (angle <26°) infrabony defects exhibited more favourable CAL gain than wide defects (not statistically significant) and statistically significantly more favourable bony gain (p < 0.05) Linares et al., 2006 GTR (collagen membrane)/deproteinized Radiographic defect angle did not have a statistically significant bovine bone mineral vs. OFD effect on CAL gain (p-value = 0.8138). Radiographic defect angle did not have statistically significant effect on bone gain either (p-value = 0.6179)

adjunctive bone replacement grafts) or GTR. On the other hand, the 56% (Christgau et al., 2002; Falk et al., 1997). Furthermore, a 1-wall more clinically meaningful CAL outcome was not associated with component may be present in the majority of sites (P. Cortellini & baseline defect depth. Tonetti, 2011). A gradient effect on percentage of defect fill has Low level of evidence suggests that narrower angles are associ- been shown from the 3-wall component of the defects (95 ± 6.2%) ated with more radiographic bone gain (only at categorical analysis to the 2-wall component (82 ± 18.7%) and the 1-wall component with 37° threshold) and with more CAL gain (magnitude approx- (39 ± 62.4%) in a study using GTR (Cortellini et al., 1993). In con- imately 1 mm more CAL gain for angle <37°). Furthermore, more trast, other researchers reported limited influence of the defect's walls are associated with more radiographic bone gain and CAL gain characteristics on the clinical outcome as defect characteristics (magnitude approximately 0.5 mm per extra wall from 1 to 2 to 3). showed weak or no correlations to defect fill (Polson & Heijl, 1978; The increased chance of favourable treatment outcomes following Renvert et al., 1985). It is clear that defect depth, narrow angle and periodontal surgery by number of residual walls has also been widely increased number of walls are correlated, since usually the deepest reported in publications not included in this review (Rosling et al., part of the defect has increased walls and it is narrower. Therefore, 1976). When categorizing infrabony defects by number of walls, it might be difficult to disentangle the relative contribution of each we should not forget that often defects are categorized as “combi- of these morphology aspects on regenerative surgery outcomes. It is nations” and relative proportions of 1-, 2- or 3-wall components of also important to highlight that these observations, based on studies the defects are reported (Cortellini et al., 2008). The prevalence of inclusion, were specific to defects with at least 3-mm radiographic “combination” defects in some GTR studies was as high as 31% to infrabony component. These findings are further confounded by the 108 | NIBALI et al.

(A)

(B)

(C)

FIGURE 2 Forest plots of meta-analysis of effect of defect angle on healing following regenerative surgery: categorical analysis for the effect of initial defect angle <37° on radiographic bone gain at 12 months (2A), regression estimates for the effect of initial defect angle on radiographic bone gain at 12 months (2B) and regression estimates for the effect of defect angle on CAL gain at 12 months (2C)

fact that, in combination defects, the subcomponents with a lower defects extended to buccal and/or lingual walls, with the exception number of walls are the more superficial ones, which may be neg- of a few studies including circumferential defects (Al Machot et al., atively affected by the oral environment and the wound-healing 2014; Hoffmann et al., 2016; Meyle et al., 2011). Such extension is process. often pivotal for decision-making on biomaterials to be used and Another important finding of this review is that, although most on flap design (e.g. MIST vs. M-MIST or single-flap approach). No publications reported some description of the study defects, only a mention of “craters” was found in the reviewed papers, although minority of publications did report outcomes based on defect mor- other papers not included in this review have described attempted phology. This is somehow surprising, since it has long been suspected regeneration of this type of defects (Falk et al., 1997). Therefore, no that regeneration of intra-osseous defects is thought to depend on meaningful conclusion can be drawn on the regenerative potential of uninterrupted maturation of the fibrin clot, favoured by stability of craters and on the potential effect of defect extension on outcomes the wound and good soft tissue coverage of the defect (Hiatt et al., of regenerative therapy. 1968; Wikesjo & Nilveus, 1990). As such, intra-osseous defect mor- Different types of regenerative materials can be adapted to the phology is believed to be a crucial factor to facilitate predictable defect morphology: some materials are supportive and space-main- regeneration by influencing stability of the blood clot. These initial taining, such as non-resorbable membranes, bone grafts and com- theories are supported by observations in animal models, showing bination of resorbable membrane and bone grafts, and others are that bone and cementum regeneration was positively correlated to non-supportive and non-space-maintaining materials, such as resorb- the number of bone walls limiting the infrabony periodontal defects able membranes alone, enamel matrix proteins and growth factors (Kim et al., 2004). Interestingly, the description of defect morphol- (Pierpaolo Cortellini & Tonetti, 2015). Some publications have sug- ogy in the included papers was limited to depth, angle and number gested that supportive biomaterials may overcome the negative ef- of residual walls in the inter-proximal area and not to whether the fect of defect morphology and improve the outcomes of regeneration NIBALI et al. | 109

TABLE 3 Details of papers reporting treatment outcomes divided by baseline defect morphology

Authors Defects included Regenerative treatment provided Comparison result by defect type

Cortellini et al., 1-, 2-, 3-wall and MIST+EMD No statistically significant association for CAL 2008 combined change at multilevel regression analysis for 3-wall vs. other defects at 12 months (p = 0.135) Cortellini et al., 1-, 2-, 3-wall and Gore-Tex Bone gain at 12 months: 3-wall = 2.7 ± 2.2, 1993 combined 2-wall = 1.6 ± 1.6, 1-wall = 0.4 ± 1.6 (bone fill: 95 ± 6.2%, 82 ± 18.7% and 39 ± 62.4% for 3-, 2- and 1-wall components, respectively) Cosyn et al., 1-, 2-, 3-wall and M-MIST vs. MIST+Bio-Oss Association between 1-wall defects and both 2012 combined failure (CAL gain <2 mm at 12 months, OR = 10.4 for 1-wall vs. 2-wall defects) and increased buccal recession (OR = 58.8 for 1- vs. 2-wall defects) Crea et al., 2-, 3-wall or OFD+intra-marrow penetration IMP vs. OFD Radiographic defect depth change: 2-wall OFD 2014 combined (n = 4):1.00 ± 1.82, OFD+IMP (n = 13): 3.14 ± 1.36, 3-wall OFD (n = 9): 2.00 ± 1.12, OFD+IMP (n = 15): 3.00 ± 1.76. CAL change: 2-wall OFD (n = 4):1.75 ± 3.33, OFD+IMP (n = 13): 3.00 ± 1.62, 3-wall OFD (n = 9): 1.78 ± 2.63, OFD+IMP (n = 15): 3.14 ± 1.85 Loos et al., 1-, 2- or 3-wall Bioresorbable membrane or not (with or without No statistically significant association between 2002 systemic antibiotics) number of walls and CAL gain Losada et al., 1-, 2 wall or EMD+BC vs. EMD Probability of gaining ≥3 mm CAL: 2.57 (0.36- 2017 combined 18.33) times higher for narrow defect angles (<24.75°) and 0.55 (0.16-1.92) for 1-wall vs. 2-wall Meyle et al., 1-, 2-wall, combined EMD +synthetic bone graft vs. EMD Tukey's plot showing bone fill in each defect type 2011 1- and 2-wall, and circumferential Silvestri et al., 2-, 3-wall EMD vs. GTR CAL gain: 3-wall = 5.02 ± 1.86 mm, 2003 2-wall = 4.25 ± 2.34 mm Stavropoulos 1-, 2-wall and GTR (resorbable membrane) vs. resorbable Estimated differences for 1-wall vs. 2-wall: PPD et al., 2003 combined 1- and membrane +Bio-Oss impregnated with saline gain (0.40 mm, CI = −0.32;1.12, p = 0.26), CAL 2-wall vs. resorbable membrane +with Bio- gain (0.44 mm, CI = 1.68; 0.74, p = 0.44) and Oss impregnated with gentamicin vs. flap surgery bone gain (1.33 mm, CI = 0.14; 2.53, p = 0.03) Tonetti et al., 1-, 2- and 3-wall Papilla preservation flap with EMD vs. no OR of CAL gain ≥3 mm: 3-wall vs. 1-wall: 2.69 2002 regenerative material (CI = 1.1-7.5) Tonetti et al., 1-, 2- and 3-wall Papilla preservation flap with GTR vs. no Estimated difference in CAL gain in 1- vs. 2004 regenerative material 3-wall = −0.5 ± 0.04 Yukna et al. 1-, 2-, 3-wall and Calcium carbonate graft vs. OFD Relative defect fill number (≥90%, ≥50%, <50%, 1994 combined <10% of defect) presented for each defect type Tonetti, et al., 1-, 2-, 3-wall and GTR Titanium ePTFE vs. GTR ePTFE vs. OFD General linear model showing lack of significance 1996 combined of depth of 1-, 2- or 3-wall on 1-year CAL gain (p-value = 0.664, 0.24 and 0.19, respectively)

Nevins et al., 1,2 wall, β-TCP +sodium acetate (SA) vs. β-TCP +SA Graph showing the influence of defect type on 2013 combined or +0.3 mg/mL rhPDGF-BB vs. β-TCP +SA CAL gain with time circumferential +1.0 mg/mL rhPDGF-BB

in non-space-maintaining defects as they have the ability to create such as EMD, their added benefit is thought to be greater in defects and maintain space for regeneration and provide increased stability with a predominantly 3-wall anatomy compared with one-wall defect to the blood clot (Palmer, Cortellini, & Group, 2008; Reynolds et al., (Tonetti et al., 2002). However, these concepts are in constant evo- 2003; Slotte et al., 2012; Tonetti et al., 1993, 1996, 2004; Trombelli lution together with developments in the surgical procedure itself, & Farina, 2008). On the other hand, for non-supportive biomaterials, specifically flap design and suturing technique. The most common 110 | NIBALI et al.

(A)

(B)

(C)

(D)

FIGURE 3 Forest plots of meta-analysis of effect of number of walls on healing following regenerative surgery: categorical analysis for the effect of 1-wall vs. 2-wall defects on radiographic bone gain at 12 months (3A), categorical analysis for the effect of 2-wall vs. 3-wall on radiographic bone gain at 12 months (3B), regression estimates for the effect of number of walls on radiographic bone gain at 12 months (3C) and regression estimates for the effect of number of walls on CAL gain at 12 months (3D)

complications of periodontal regeneration procedures were dehis- healing phase. It has been shown that stable flap designs, such as cence of inter-dental tissues, graft exfoliation, membrane collapse achieved by minimally invasive surgical therapy (MIST), can lead to and/or exposure, with the subsequent bacterial contamination which such favourable regenerative outcomes that the use of regenera- negatively affects the outcomes of regeneration, such as CAL gain tive materials may not offer any additional benefits (Liu et al., 2016). and bone gain (P. Cortellini et al., 1993; P. Cortellini et al., 2001; Selvig There is no clear effect of biomaterials (supportive or not) on these et al., 1992). Therefore, new surgical techniques were developed with results, although the only factor where this could be formally anal- the aim of soft tissue preservation in order to achieve tension-free ysed was defect depth. primary closure over the defect and the regenerative materials, and A recent systematic review has concluded that EMD and GTR to ensure wound stability and blood clot stability during the early with resorbable membranes appear to be the gold standard for the NIBALI et al. | 111

surgical treatment of deep (≥3 mm) infrabony defects which have not resolved following completion of non-surgical therapy and that among the possible replacement biomaterials, deproteinized bovine bone mineral (DBBM) improved clinical outcomes of both EMD and resorbable GTR compared with OFD and it should be considered a Strength of evidence LOW HIGH LOW LOW–MODERATE MODERATE LOW - MODERATE–LOW viable treatment option especially in non-supporting defects (Nibali et al., 2019). These authors also suggested that papillary preservation flaps may improve the clinical outcomes and should be considered a surgical pre-requisite when performing any regeneration procedure (Nibali et al., 2019). These observations are the basis for the recently published EFP S3 clinical guidelines (Sanz et al., 2020). - Number of studies with low RoB 4 - 2 - - - 3 The topic of periodontal regeneration of periodontal infraosseous defects is developing quickly with other biologically active agents such as growth factors (Smith et al., 2015) and bone morphogenetic proteins (Larsson et al., 2016). Therefore, the frontier of what is “re-

) generable” is quickly moving and the “bar” is being raised. 2 I - Heterogeneity Heterogeneity ( 4% - 33% - - - 80% Strengths of this review are the analysis of a large body of literature and the relatively low heterogeneity, leading to moderate-to-high strength of evidence for most meta-analyses. The exclusion of one paper (Cosyn et al., 2012) significantly reduced heterogeneity for the radiographic bone gain outcomes. Risk of bias revealed that only a minority of papers were defined as “poor quality” across all study designs, with “reporting bias” for RCT and “compa- risk of bias. Regression analyses were carried out having outcomes (bone gain and

No data CAL gain Effect size if statisticallysignificant NS - OR 0.97 No data No data No data OR 1.42 = rability” for cohort studies resulting as areas requiring improvement. The main limitation of this review is that despite the inclusion of more than a hundred papers, only 15 papers could be included in meta-analyses, due to limited or heterogeneous data reporting. From these data, it emerges clearly how infrabony defect morphology has an important influence on outcomes of regenerative 3 Number of studies with low RoB 6 2 2 2 1 0 2 periodontal surgery. Baseline defect depth seems to positively influence radiographic bone gain 12 months post-surgery, while narrower angles and increased number of walls positively influence both bone not statistically significant, RoB

= and CAL gain. A good description and definition of the infraosseous

) 2

I defects can help in planning the most appropriate treatment option. 52% Heterogeneity Heterogeneity ( 0% 50% 80% 0% 54% - 0% Such specific definition can only really be obtained intra-surgically or perhaps through CBCT scan, although combined accurate assess-

odds ratio, NS ment of probing pocket depths and periapical radiographs has good

= value (Wolf et al., 2001) and should be sufficient for treatment plan- m m m ning of most cases. A detailed classification system for infraosseous m m m m m m

defects, which takes into account also other aspects of defect morphology, such as extension to buccal–lingual walls, should be used −0.75 Effect size if statisticallysignificant Radiographic bone gain bone Radiographic OR 1.32 0.94 NS −0.57 −0.39 −1.18 OR 3.43 widely to improve our understanding of regenerative potential and

m) of appropriate biomaterials for different types of defects. m

ORCID Luigi Nibali https://orcid.org/0000-0002-7750-5010 George Pelekos https://orcid.org/0000-0003-1917-1988 Threshold (<4 Regression analysis Regression Threshold 37°) Threshold Regression analysis Regression 1 vs. 2 2 vs. 3 1 vs. 3 Regression analysis Regression Guo-Hao Lin https://orcid.org/0000-0003-1290-9994

REFERENCES Al Machot, E., Hoffmann, T., Lorenz, K., Khalili, I., & Noack, B. (2014). Clinical outcomes after treatment of periodontal intrabony de- L gain) as continuous variables Defect depth Morphology parameter Morphology Defect angle Number of walls fects with nanocrystalline hydroxyapatite (Ostim) or enamel CA TABLE 4 SummaryTABLE of all meta-analyses findings. OR 112 | NIBALI et al.

matrix derivatives (Emdogain): a randomized controlled clinical Crea, A., Deli, G., Littarru, C., Lajolo, C., Orgeas, G. V., & Tatakis, D. N. trial. BioMed Research International, 2014, 786353. https://doi. (2014). Intrabony defects, open-flap debridement, and decortica- org/10.1155/2014/786353 tion: a randomized clinical trial. Journal of Periodontology, 85(1), 34- Bratthall, G., Lindberg, P., Havemose-Poulsen, A., Holmstrup, P., Bay, L., 42. https://doi.org/10.1902/jop.2013.120753 Soderholm, G., & Skold Bell, H. (2001). Comparison of ready-to- Ehmke, B., Rudiger, S. G., Hommens, A., Karch, H., & Flemmig, T. F. use EMDOGAIN-gel and EMDOGAIN in patients with chronic adult (2003). Guided tissue regeneration using a polylactic acid bar- periodontitis. Journal of Clinical Periodontology, 28(10), 923-929. rier. Journal of Clinical Periodontology, 30(4), 368-374. https://doi. https://doi.org/10.1034/j.1600-051x.2001.028010923.x org/10.1034/j.1600-051x.2003.00312.x Briguglio, F., Briguglio, E., Briguglio, R., Cafiero, C., & Isola, G. (2013). Eickholz, P., Horr, T., Klein, F., Hassfeld, S., & Kim, T. S. (2004). Treatment of infrabony periodontal defects using a resorbable bio- Radiographic parameters for prognosis of periodontal healing polymer of hyaluronic acid: a randomized clinical trial. Quintessence of infrabony defects: two different definitions of defect depth. International, 44(3), 231-240. https://doi.org/10.3290/j.qi.a29054 Journal of Periodontology, 75(3), 399-407. https://doi.org/10.1902/ Castro, A. B., Meschi, N., Temmerman, A., Pinto, N., Lambrechts, P., jop.2004.75.3.399 Teughels, W., & Quirynen, M. (2017). Regenerative potential of Eickholz, P., Röllke, L., Schacher, B., Wohlfeil, M., Dannewitz, B., leucocyte- and platelet-rich fibrin. Part A: intra-bony defects, fur- Kaltschmitt, J., Krieger, J. K., Krigar, D. M., Reitmeir, P., & Kim, T.- cation defects and periodontal plastic surgery. A systematic review S. (2014). Enamel matrix derivative in propylene glycol alginate for and meta-analysis. Journal of Clinical Periodontology, 44(1), 67-82. treatment of infrabony defects with or without systemic doxycy- https://doi.org/10.1111/jcpe.12643 cline: 12- and 24-month results. Journal of Periodontology, 85(5), Christgau, M., Bader, N., Felden, A., Gradl, J., Wenzel, A., & Schmalz, 669-675. https://doi.org/10.1902/jop.2013.130290 G. (2002). Guided tissue regeneration in intrabony defects using Ellegaard, B., & Loe, H. (1971). New attachment of periodontal tissues an experimental bioresorbable polydioxanon (PDS) membrane. after treatment of intrabony lesions. Journal of Periodontology, A 24-month split-mouth study. Journal of Clinical Periodontology, 42(10), 648-652. https://doi.org/10.1902/jop.1971.42.10.648 29(8), 710-723. https://doi.org/10.1034/j.1600-051x.2002. Falk, H., Laurell, L., Ravald, N., Teiwik, A., & Persson, R. (1997). Guided 290808.x tissue regeneration therapy of 203 consecutively treated intra- Cortellini, P., Nieri, M., Prato, G. P., & Tonetti, M. S. (2008). Single mini- bony defects using a bioabsorbable matrix barrier. Clinical and mally invasive surgical technique with an enamel matrix derivative Radiographic Findings. Journal of Periodontology, 68(6), 571-581. to treat multiple adjacent intra-bony defects: clinical outcomes and https://doi.org/10.1902/jop.1997.68.6.571 patient morbidity. Journal of Clinical Periodontology, 35(7), 605-613. Francetti, L., Trombelli, L., Lombardo, G., Guida, L., Cafiero, C., Roccuzzo, https://doi.org/10.1111/j.1600-051X.2008.01242.x M., & Del Fabbro, M. (2005). Evaluation of efficacy of enamel ma- Cortellini, P., Pini Prato, G., & Tonetti, M. S. (1993). Periodontal regen- trix derivative in the treatment of intrabony defects: a 24-month eration of human infrabony defects. II. Re-entry procedures and multicenter study. International Journal of Periodontics & Restorative bone measures. Journal of Periodontology, 64(4), 261-268. https:// Dentistry, 25(5), 461-473. doi.org/10.1902/jop.1993.64.4.261 Goldman, H. M., & Cohen, D. W. (1958). The Infrabony Pocket: Cortellini, P., Stalpers, G., Mollo, A., & Tonetti, M. S. (2020). Periodontal Classification and Treatment. The Journal of Periodontology, 29(4), regeneration versus extraction and dental implant or prosthetic re- 272-291. https://doi.org/10.1902/jop.1958.29.4.272 placement of teeth severely compromised by attachment loss to Grusovin, M. G., & Esposito, M. (2009). The efficacy of enamel matrix de- the apex: A randomized controlled clinical trial reporting 10-year rivative (Emdogain) for the treatment of deep infrabony periodon- outcomes, survival analysis and mean cumulative cost of recur- tal defects: a placebo-controlled randomised clinical trial. European rence. Journal of Clinical Periodontology, 47(6), 768-776. https://doi. Journal of Oral Implantology, 2(1), 43-54. org/10.1111/jcpe.13289 Guyatt, G. H., Oxman, A. D., Vist, G. E., Kunz, R., Falck-Ytter, Y., Alonso- Cortellini, P., & Tonetti, M. S. (2011). Clinical and radiographic outcomes Coello, P., & Schünemann, H. J.; Group, G. W. (2008). GRADE: an of the modified minimally invasive surgical technique with and emerging consensus on rating quality of evidence and strength without regenerative materials: a randomized-controlled trial in in- of recommendations. BMJ, 336(7650), 924-926. https://doi. tra-bony defects. Journal of Clinical Periodontology, 38(4), 365-373. org/10.1136/bmj.39489.470347.AD https://doi.org/10.1111/j.1600-051X.2011.01705.x Heins, P., Hartigan, M., Low, S., & Chace, R. (1989). Relative stability of Cortellini, P., & Tonetti, M. S. (2015). Clinical concepts for regenerative deep- versus shallow-side bone levels in angular proximal infrabony therapy in intrabony defects. Periodontology 2000, 68(1), 282–307. defects. Journal of Clinical Periodontology, 16(1), 59-64. https://doi. https://doi.org/10.1111/prd.12048 org/10.1111/j.1600-051x.1989.tb01613.x Cortellini, P., Tonetti, M. S., Lang, N. P., Suvan, J. E., Zucchelli, G., Vangsted, Hiatt, W. H., Stallard, R. E., Butler, E. D., & Badgett, B. (1968). Repair T., Silvestri, M., Rossi, R., McClain, P., Fonzar, A., Dubravec, D., & following mucoperiosteal flap surgery with full gingival retention. Adriaens, P. (2001). The simplified papilla preservation flap in the Journal of Periodontology, 39(1), 11-16. https://doi.org/10.1902/ regenerative treatment of deep intrabony defects: clinical out- jop.1968.39.1.11 comes and postoperative morbidity. Journal of Periodontology, Hoffmann, T., Al-Machot, E., Meyle, J., Jervoe-Storm, P. M., & Jepsen, S. 72(12), 1702-1712. https://doi.org/10.1902/jop.2001.72.12.1702 (2016). Three-year results following regenerative periodontal sur- Cosyn, J., Cleymaet, R., Hanselaer, L., & De Bruyn, H. (2012). gery of advanced intrabony defects with enamel matrix derivative Regenerative periodontal therapy of infrabony defects using min- alone or combined with a synthetic bone graft. Clin Oral Investig, imally invasive surgery and a collagen-enriched bovine-derived 20(2), 357-364. https://doi.org/10.1007/s00784-015-1522-4 xenograft: a 1-year prospective study on clinical and aesthetic out- Ilgenli, T., Dundar, N., & Kal, B. I. (2007). Demineralized freeze-dried come. Journal of Clinical Periodontology, 39(10), 979-986. https:// bone allograft and platelet-rich plasma vs platelet-rich plasma alone doi.org/10.1111/j.1600-051X.2012.01924.x in infrabony defects: a clinical and radiographic evaluation. Clin Oral Crea, A., Dassatti, L., Hoffmann, O., Zafiropoulos, G. G., & Deli, G. (2008). Investig, 11(1), 51-59. https://doi.org/10.1007/s00784-006-0083-y Treatment of intrabony defects using guided tissue regeneration Karn, K. W., Shockett, H. P., Moffitt, W. C., & Gray, J. L. (1984). or enamel matrix derivative: a 3-year prospective randomized clin- Topographic classification of deformities of the alveolar process. ical study. Journal of Periodontology, 79(12), 2281-2289. https://doi. Journal of Periodontology, 55(6), 336-340. https://doi.org/10.1902/ org/10.1902/jop.2008.080135 jop.1984.55.6.336 NIBALI et al. | 113

Kim, C. S., Choi, S. H., Chai, J. K., Cho, K. S., Moon, I. S., Wikesjo, U. M., & Nibali, L., Koidou, V. P., Nieri, M., Barbato, L., Pagliaro, U., & Cairo, F. Kim, C. K. (2004). Periodontal repair in surgically created intrabony (2019). Regenerative surgery versus access flap for the treatment defects in dogs: influence of the number of bone walls on healing of intrabony periodontal defects. A systematic review and me- response. Journal of Periodontology, 75(2), 229-235. https://doi. ta-analysis. Journal of Clinical Periodontology, 47(S22), 320–351. org/10.1902/jop.2004.75.2.229 https://doi.org/10.1111/jcpe.13237 Klein, F., Kim, T. S., Hassfeld, S., Staehle, H. J., Reitmeir, P., Holle, R., Nyman, S., Lindhe, J., Karring, T., & Rylander, H. (1982). New attachment & Eickholz, P. (2001). Radiographic defect depth and width for following surgical treatment of human periodontal disease. Journal prognosis and description of periodontal healing of infrabony de- of Clinical Periodontology, 9, 290-296. fects. Journal of Periodontology, 72(12), 1639-1646. https://doi. Palmer, R. M., & Cortellini, P.; Group, B. o. E. W. o. P. (2008). org/10.1902/jop.2001.72.12.1639 Periodontal tissue engineering and regeneration: Consensus Larsson, L., Decker, A. M., Nibali, L., Pilipchuk, S. P., Berglundh, T., & Report of the Sixth European Workshop on Periodontology. Giannobile, W. V. (2016). Regenerative Medicine for Periodontal Journal of Clinical Periodontology, 35(8 Suppl), 83-86. https://doi. and Peri-implant Diseases. Journal of Dental Research, 95(3), 255- org/10.1111/j.1600-051X.2008.01262.x 266. https://doi.org/10.1177/0022034515618887 Papapanou, P. N., & Tonetti, M. S. (2000). Diagnosis and epidemiology Liberati, A., Altman, D. G., Tetzlaff, J., Mulrow, C., Gøtzsche, P. C., of periodontal osseous lesions. Periodontol, 2000(22), 8-21. https:// Ioannidis, J. P. A., Clarke, M., Devereaux, P. J., Kleijnen, J., & Moher, doi.org/10.1034/j.1600-0757.2000.2220102.x D. (2009). The PRISMA statement for reporting systematic reviews Papapanou, P. N., & Wennstrom, J. L. (1991). The angular bony de- and meta-analyses of studies that evaluate health care interven- fect as indicator of further alveolar bone loss. Journal of Clinical tions: explanation and elaboration. Journal of Clinical Epidemiology, Periodontology, 18(5), 317-322. https://doi.org/10.1111/j.1600- 62(10), e1-34. https://doi.org/10.1016/j.jclinepi.2009.06.006 051x.1991.tb00435.x Linares, A., Cortellini, P., Lang, N. P., Suvan, J., & Tonetti, M. S. & European Polson, A. M., & Heijl, L. C. (1978). Osseous repair in infrabony periodon- Research Group on, P. (2006). Guided tissue regeneration/depro- tal defects. Journal of Clinical Periodontology, 5(1), 13-23. https:// teinized bovine bone mineral or papilla preservation flaps alone for doi.org/10.1111/j.1600-051x.1978.tb01902.x treatment of intrabony defects. II: radiographic predictors and out- Pontoriero, R., Nyman, S., & Lindhe, J. (1988). The angular bony defect comes. Journal of Clinical Periodontology, 33(5), 351–358. https:// in the maintenance of the periodontal patient. Journal of Clinical doi.org/10.1111/j.1600-051X.2006.00911.x Periodontology, 15(3), 200-204. https://doi.org/10.1111/j.1600- Liu, S., Hu, B., Zhang, Y., Li, W., & Song, J. (2016). Minimally Invasive 051x.1988.tb01570.x Surgery Combined with Regenerative Biomaterials in Treating Renvert, S., Garrett, S., Nilveus, R., Chamberlain, A. D., & Egelberg, J. Intra-Bony Defects: A Meta-Analysis. PLoS One, 11(1), e0147001. (1985). Healing after treatment of periodontal intraosseous de- https://doi.org/10.1371/journal.pone.0147001 fects. VI. Factors influencing the healing response. Journal of Loos, B. G., Louwerse, P. H., Van Winkelhoff, A. J., Burger, W., Gilijamse, Clinical Periodontology, 12(9), 707-715. https://doi.org/10.1111/ M., Hart, A. A., & van der Velden, U. (2002). Use of barrier mem- j.1600-051x.1985.tb01396.x branes and systemic antibiotics in the treatment of intraosseous Reynolds, M. A., Aichelmann-Reidy, M. E., Branch-Mays, G. L., & defects. Journal of Clinical Periodontology, 29(10), 910-921. https:// Gunsolley, J. C. (2003). The efficacy of bone replacement grafts in doi.org/10.1034/j.1600-051x.2002.291006.x the treatment of periodontal osseous defects. A Systematic review. Losada, M., Gonzalez, R., Garcia, A. P., Santos, A., & Nart, J. (2017). Annals of Periodontology, 8(1), 227-265. https://doi.org/10.1902/ Treatment of Non-Contained Infrabony Defects With Enamel annals.2003.8.1.227 Matrix Derivative Alone or in Combination With Biphasic Calcium Rosling, B., Nyman, S., Lindhe, J., & Jern, B. (1976). The healing poten- Phosphate Bone Graft: A 12-Month Randomized Controlled tial of the periodontal tissues following different techniques of Clinical Trial. Journal of Periodontology, 88(5), 426-435. https://doi. periodontal surgery in plaque-free dentitions. A 2-year clinical org/10.1902/jop.2016.160459 study. Journal of Clinical Periodontology, 3(4), 233-250. https://doi. Melcher, A. H. (1976). On the repair potential of periodontal tissues. org/10.1111/j.1600-051x.1976.tb00042.x Journal of Periodontology, 47, 256-260. Sanz, M., Herrera, D., Kebschull, M., Chapple, I., Jepsen, S., Beglundh, T., Meyle, J., Hoffmann, T., Topoll, H., Heinz, B., Al-Machot, E., Jervøe- Sculean, A., Tonetti, M. S., Merete Aass, A., Aimetti, M., Kuru, B. E., Storm, P.-M., Meiß, C., Eickholz, P., & Jepsen, S. (2011). A multi-cen- Belibasakis, G., Blanco, J., Bol-van den Hil, E., Bostanci, N., Bozic, tre randomized controlled clinical trial on the treatment of in- D., Bouchard, P., Buduneli, N., Cairo, F., … Wennström, J. (2020). tra-bony defects with enamel matrix derivatives/synthetic bone Treatment of Stage I-III Periodontitis -The EFP S3 Level Clinical graft or enamel matrix derivatives alone: results after 12 months. Practice Guideline. Journal of Clinical Periodontology, https://doi. Journal of Clinical Periodontology, 38(7), 652-660. https://doi. org/10.1111/jcpe.13290 org/10.1111/j.1600-051X.2011.01726.x Sanz, M., Tonetti, M. S., Zabalegui, I., Sicilia, A., Blanco, J., Rebelo, Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Preferred H., Rasperini, G., Merli, M., Cortellini, P., & Suvan, J. E. (2004). reporting items for systematic reviews and meta-analyses: the Treatment of intrabony defects with enamel matrix proteins or PRISMA statement. Journal of Clinical Epidemiology, 62(10), 1006- barrier membranes: results from a multicenter practice-based 1012. https://doi.org/10.1016/j.jclinepi.2009.06.005 clinical trial. Journal of Periodontology, 75(5), 726-733. https://doi. Needleman, I. G., Worthington, H. V., Giedrys-Leeper, E., & Tucker, R. org/10.1902/jop.2004.75.5.726 J. (2006). Guided tissue regeneration for periodontal infra-bony Selvig, K. A., Kersten, B. G., Chamberlain, A. D., Wikesjo, U. M., & Nilveus, R. defects. Cochrane Database of Systematic Reviews, https://doi. E. (1992). Regenerative surgery of intrabony periodontal defects using org/10.1002/14651858.CD001724.pub2 ePTFE barrier membranes: scanning electron microscopic evaluation of Nevins, M., Kao, R. T., McGuire, M. K., McClain, P. K., Hinrichs, J. E., retrieved membranes versus clinical healing. Journal of Periodontology, McAllister, B. S., Reddy, M. S., Nevins, M. L., Genco, R. J., Lynch, 63(12), 974-978. https://doi.org/10.1902/jop.1992.63.12.974 S. E., & Giannobile, W. V. (2013). Platelet-derived growth factor Selvig, K. A., Kersten, B. G., & Wikesjo, U. M. (1993). Surgical treatment promotes periodontal regeneration in localized osseous defects: of intrabony periodontal defects using expanded polytetrafluoro- 36-month extension results from a randomized, controlled, dou- ethylene barrier membranes: influence of defect configuration on ble-masked clinical trial. Journal of Periodontology, 84(4), 456-464. healing response. Journal of Periodontology, 64(8), 730-733. https:// https://doi.org/10.1902/jop.2012.120141 doi.org/10.1902/jop.1993.64.8.730 114 | NIBALI et al.

Silvestri, M., Sartori, S., Rasperini, G., Ricci, G., Rota, C., & Cattaneo, V. Tsitoura, E., Tucker, R., Suvan, J., Laurell, L., Cortellini, P., & Tonetti, M. (2003). Comparison of infrabony defects treated with enamel ma- (2004). Baseline radiographic defect angle of the intrabony defect trix derivative versus guided tissue regeneration with a nonresorb- as a prognostic indicator in regenerative periodontal surgery with able membrane. Journal of Clinical Periodontology, 30(5), 386-393. enamel matrix derivative. Journal of Clinical Periodontology, 31(8), https://doi.org/10.1034/j.1600-051x.2003.10146.x 643-647. https://doi.org/10.1111/j.1600-051X.2004.00555.x Slotte, C., Asklow, B., Sultan, J., & Norderyd, O. (2012). A random- Weinberg, M. A., & Eskow, R. N. (2000). Osseous defects: proper termi- ized study of open-flap surgery of 32 intrabony defects with nology revisited. Journal of Periodontology, 71(12), 1928. https://doi. and without adjunct bovine bone mineral treatment. Journal org/10.1902/jop.2000.71.12.1928 of Periodontology, 83(8), 999-1007. https://doi.org/10.1902/ Wikesjo, U. M., & Nilveus, R. (1990). Periodontal repair in dogs: effect jop.2011.110490 of wound stabilization on healing. Journal of Periodontology, 61(12), Smith, P. C., Martínez, C., Cáceres, M., & Martínez, J. (2015). Research 719-724. https://doi.org/10.1902/jop.1990.61.12.719 on growth factors in periodontology. Periodontology 2000, 67(1), Wolf, B., von Bethlenfalvy, E., Hassfeld, S., Staehle, H. J., & Eickholz, P. 234–250. https://doi.org/10.1111/prd.12068 (2001). Reliability of assessing interproximal bone loss by digital ra- Stavropoulos, A., Karring, E. S., Kostopoulos, L., & Karring, T. (2003). diography: intrabony defects. Journal of Clinical Periodontology, 28(9), Deproteinized bovine bone and gentamicin as an adjunct to GTR in 869-878. https://doi.org/10.1034/j.1600-051x.2001.028009869.x the treatment of intrabony defects: a randomized controlled clini- Xu, Y., Qiu, J., Sun, Q., Yan, S., Wang, W., Yang, P., & Song, A. (2019). cal study. Journal of Clinical Periodontology, 30(6), 486-495. https:// One-Year Results Evaluating the Effects of Concentrated Growth doi.org/10.1034/j.1600-051x.2003.00258.x Factors on the Healing of Intrabony Defects Treated with or Tonetti, M. S., Cortellini, P., Lang, N. P., Suvan, J. E., Adriaens, P., Dubravec, without Bone Substitute in Chronic Periodontitis. Medical Science D., Fonzar, A., Fourmousis, I., Rasperini, G., Rossi, R., Silvestri, M., Monitor, 25, 4384-4389. https://doi.org/10.12659/MSM.917025 Topoll, H., Wallkamm, B., & Zybutz, M. (2004). Clinical outcomes Zucchelli, G., Amore, C., Montebugnoli, L., & De Sanctis, M. (2003). following treatment of human intrabony defects with GTR/bone Enamel matrix proteins and bovine porous bone mineral in the replacement material or access flap alone. A multicenter random- treatment of intrabony defects: a comparative controlled clini- ized controlled clinical trial. Journal of Clinical Periodontology, 31(9), cal trial. Journal of Periodontology, 74(12), 1725-1735. https://doi. 770-776. https://doi.org/10.1111/j.1600-051X.2004.00562.x org/10.1902/jop.2003.74.12.1725 Tonetti, M. S., Lang, N. P., Cortellini, P., Suvan, J. E., Adriaens, P., Zucchelli, G., Bernardi, F., Montebugnoli, L., & De Sanctis, M. (2002). Dubravec, D., Fonzar, A., Fourmousis, I., Mayfield, L., Rossi, R., Enamel Matrix Proteins and Guided Tissue Regeneration Silvestri, M., Tiedemann, C., Topoll, H., Vangsted, T., & Wallkamm, With Titanium-Reinforced Expanded Polytetrafluoroethylene B. (2002). Enamel matrix proteins in the regenerative therapy of Membranes in the Treatment of Infrabony Defects: A Comparative deep intrabony defects. Journal of Clinical Periodontology, 29(4), Controlled Clinical Trial. Journal of Periodontology, 73(1), 3-12. 317-325. https://doi.org/10.1034/j.1600-051x.2002.290407.x https://doi.org/10.1902/jop.2002.73.1.3 Tonetti, M. S., Pini Prato, G., Stalpers, G., & Cortellini, P. (1996). Guided tissue regeneration of deep intrabony defects in strategically important prosthetic abutments. International Journal of Periodontics & Restorative Dentistry, 16(4), 378-387. SUPPORTING INFORMATION Tonetti, M. S., Pini-Prato, G., & Cortellini, P. (1993). Periodontal regen- Additional supporting information may be found online in the eration of human intrabony defects. IV. Determinants of healing Supporting Information section. response. Journal of Periodontology, 64(10), 934-940. https://doi. org/10.1902/jop.1993.64.10.934 Tonetti, M. S., Prato, G. P., & Cortellini, P. (1996). Factors affecting the How to cite this article: Nibali L, Sultan D, Arena C, Pelekos G, healing response of intrabony defects following guided tissue regeneration and access flap surgery. Journal of Clinical Periodontology, Lin G-H, Tonetti M. Periodontal infrabony defects: Systematic 23(6), 548-556. https://doi.org/10.1111/j.1600-051x.1996.tb018 review of healing by defect morphology following regenerative 23.x surgery. J Clin Periodontol. 2021;48:101–114. https://doi. Trombelli, L., & Farina, R. (2008). Clinical outcomes with bioactive agents org/10.1111/jcpe.13381 alone or in combination with grafting or guided tissue regeneration. Journal of Clinical Periodontology, 35(8 Suppl), 117-135. https://doi. org/10.1111/j.1600-051X.2008.01265.x Received: 3 October 2019 | Revised: 27 August 2020 | Accepted: 28 September 2020 DOI: 10.1111/jcpe.13383

ORIGINAL ARTICLE CLINICAL PERIODONTOLOGY

Incidence and progression of gingival recession over 4 years: A population-based longitudinal study

1Periodontology, Faculty of Dentistry, Federal University of Rio Grande do Sul, Abstract Porto Alegre, Brazil Aim: To describe changes in gingival recession (GR) at buccal and palatal sites in adults 2Preventive and Social Dentistry, Faculty of Dentistry, Federal University of Rio over an average follow-up of 4 years. Grande do Sul, Porto Alegre, Brazil Materials and methods: Baseline data were obtained from a multistage probabilistic representative sample of 1023 individuals aged ≥35 years from Porto Alegre, Brazil. Correspondence Alex N. Haas, Periodontology, Faculty Buccal and palatal/lingual GR were analysed. of Dentistry, Federal University of Rio Grande do Sul, Rua Ramiro Barcelos, 2492. Results: 402 individuals (6,862 teeth) were followed. At baseline, 3,356 (48.9%) teeth Porto Alegre, RS, Brazil. did not have GR at the buccal site and 1206 developed the condition overtime (in- Email: [email protected] cidence =35.9%; 95% CI 32.6-38.9). Percentage of incident teeth was higher among Funding information individuals with (42.3%) than those without (29.5%) periodontitis stages III/IV. Also, This study was funded by the National Coordination of Post-graduate Education 38.5% of teeth with proximal attachment loss at follow-up had incident GR compared (CAPES grant #001) and the Federal to 7.6% of those without proximal attachment loss. Incidence of palatal GR was ob- University of Rio Grande do Sul served in 32.5% of teeth (95% CI 29.7-35.3). Mean buccal and palatal/lingual GR inci (PROPESQ). - dence was 2.11 mm and 2.33 mm, whereas buccal and palatal/lingual GR progression equalled 0.40 mm and 0.48 mm. The prevalence of GR ≥3 mm increased in individuals with (from 35.9% to 47.4%) and without (from 25.2 to 41.5%) periodontitis. Conclusion: Incidence and progression of GR are high in a general urban Brazilian population of adults.

KEYWORDS Brazil, cohort, epidemiology, gingival recession

1 | INTRODUCTION Buccal GR has received great attention in clinical practice and in the scientific literature due to its important negative impacts in aes- Gingival recession (GR) is defined as the apical shift of the gingival thetics and function (Wagner et al., 2016). Periodontal plastic sur- margin caused by different conditions/pathologies (Jepsen et al., gery is predictive (Chambrone et al., 2018; Dai et al., 2019; Stefanini 2018), or even as the displacement of the gingival margin apically et al., 2018) and widely applied in clinical practice to overcome these to the enamel–cemental junction (AAP, 2001). GR is highly preva- consequences of GR at buccal sites. In this regard, the understand- lent around the world and has been associated with some risk in- ing of how buccal GR behaves over time in the community is of great dicators, such as age, oral hygiene habits and smoking (Albandar & interest. Kingman, 1999; Holtfreter et al., 2009; Löe et al., 1992; Rios et al., Some longitudinal studies showed a marked increase in buccal 2014; Sarfati et al., 2010; Susin et al., 2004). GR occurrence, extension and severity during various follow-up

wileyonlinelibrary.com/journal/jcpe | 115 J Clin Periodontol. 2021;48:115–126. 116 | RIOS et al. periods (Daprile et al., 2007; Matas et al., 2011; Serino et al., 1994; Ship & Beck, 1996). However, medium- to long-term longitudinal Clinical relevance data on incidence and progression of GR are still scarce, and this is Scientific rationale for the study: There is scarce longitudinal particularly true regarding data for adult communities (Chambrone data regarding changes in buccal and palatal/lingual gingi- & Tatakis, 2016). val recession (GR) in adult populations. The aim of this longitudinal study was to provide estimates of Principal findings: Overall proportions of teeth presenting incidence and progression of GR at buccal and palatal/lingual sites at incidence and progression of buccal GR were 35.9% and the individual level in an urban population of adults. 86.9%, respectively. However, mean incidence of buccal GR was 2.11 mm, while mean progression equalled 0.41 mm. Changes in GR differed in the presence/absence 2 | METHODS of proximal attachment loss and periodontitis. Practical implications: Preventive and therapeutic strate- 2.1 | Study design and target population gies should target sites with and without GR considering that incidence and progression estimates are high in the This population-based prospective cohort study was designed by adult population with different periodontitis stages. the Caries-Perio Collaboration Group from the Federal University of Rio Grande do Sul (UFRGS), Brazil. The Collaboration addressed a variety of oral health outcomes in the adult population from the city of Porto Alegre, studying a representative sample of men and two examiners (JG and BC), totalling four periodontists conducting women aged 35 years and older. Detailed information regarding the the examinations. At baseline and follow-up, one researcher vis- sampling strategy at baseline was previously published (Costa et al., ited each selected sector one day before the start of data collec- 2014; Rios et al., 2014; Wagner et al., 2016). In brief, a multistage tion to invite residents to participate. Residents were not included probability sampling strategy was applied. The city was divided into only after the third attempt of invitation in both examination time- 86 neighbourhoods comprising the primary sampling units that were points. Interviews and clinical examinations were conducted inside stratified in low and high income. The second stage consisted on a the household. Examinations were conducted using three portable random selection of sectors proportional to the total number of sec- devices: a medical headlight, a portable compressor and a bendable tors in each PSU. The third stage consisted of selecting households chair. consecutively according to the sector starting point until the sector sample size was reached. The number of individuals to be selected within each sector was estimated based on the proportional distri- 2.4 | Interview bution of the sample size. Baseline data were collected between June 2011 and June 2012. Participants were interviewed at baseline using a structured ques- A total of 1,600 individuals were eligible for the study at baseline tionnaire approaching demographic and behavioural information. and 1,225 individuals participated (Figure 1). Among those, 1,023 Three trained and calibrated interviewers conducted the interview. (83.5%) were dentate and were eligible for the follow-up examina- For the present study, sex, age, education, socio-economic status, tion, which was conducted between January 2016 and March 2017. smoking habits and dental care over the follow-up period were used The mean time spent (and standard deviation) between baseline and to characterize sample and population profiles. follow-up examinations was 4.2 ± 0.5 years. Age was categorized into three categories (35-49, 50-59, ≥60 years). Educational level was defined according to years of education into low (≤4 years), middle (5-10 years) and high (≥11 years). 2.2 | Ethical aspects Socio-economic status was categorized using cut-off points adapted from a Brazilian classification system (ABEP 2013) that considers the The study protocol was reviewed and obtained ethical approval from amount of consumer goods and the educational level of the head of the Research Ethics Committee, Federal University of Rio Grande do the family, as follows: low (≤20 points), middle (21-26 points) and Sul, Porto Alegre, Brazil. Prior to the interview, all patients read and high (≥27 points). signed a consent form. The total number of packs of cigarettes consumed in a lifetime (pack-years) was calculated for each individual by multiplying the number of cigarettes consumed per day by the years of habit, and di- 2.3 | Fieldwork procedures viding by 20. Smoking exposure was categorized into never-smokers (0 pack-years), light-moderate smokers (<20 pack-years) and heavy A research team of two examiners (FSR and RSC) and one assis- smokers (≥20 pack-years). tant conducted the fieldwork at baseline. At the follow-up exami- Individuals were categorized as having regular dental care if they nation, these two examiners also participated and trained other reported going to the dentist for prevention with a frequency of ≥1 RIOS et al. | 117 times/year over the follow-up period. Individuals reporting no den- midpalatal/lingual surfaces. GR was defined as the distance from the tal visits or visits only for emergencies were classified as having no cement–enamel junction (CEJ) to the free gingival margin. If the CEJ dental care. was located apical to the gingival margin, this assessment was given a negative sign. For the purposes of this study, only GR at midbuccal and midpalatal/lingual sites was included in the analyses. 2.5 | Clinical examination The examination of the major survey at baseline comprised the assessment of ten oral conditions. This included GR, which com- All permanent fully erupted teeth, excluding third molars, were ex- prised the major outcome of the survey, but not included periodon- amined using a manual periodontal probe (PCP10-SE, Hu-Friedy tal pocket depth (PPD), nor clinical attachment loss (CAL). PPD was Mfg. Co. Inc., Chicago, IL, USA). Gingival recession (GR) was assessed recorded at the follow-up examination, allowing to derive CAL for at four sites per tooth at mesiobuccal, midbuccal, distobuccal and each examined site by the sum of PPD and GR.

FIGURE 1 Flow chart of study sample 118 | RIOS et al.

2.6 | Reproducibility at the follow-up examination, applying the criteria from the 2017 World Workshop on the Classification of Periodontal and Peri-implant Reliability of the questionnaire was assessed at baseline using the Diseases and Conditions (Tonetti et al., 2018), as follows: test–retest approach in 50 participants. A set of key questions was used to assess the reproducibility of the questionnaire, and the • Stage I (at least 2 non-adjacent teeth with proximal CAL 1 to Kappa coefficients varied from 0.91 to 0.99. 2 mm); Intra- and inter-examiner reproducibility of GR was assessed be- • Stage II (at least 2 non-adjacent teeth with proximal CAL 3 to fore the start of the fieldwork at baseline and follow-up with patient 4 mm); seeking treatment in the Faculty of Dentistry at UFRGS. Duplicate • Stages III/IV (at least 2 non-adjacent teeth with proximal CAL examinations were conducted using the same equipment that would ≥5 mm). be used in the fieldwork to allow the examiners to experience the same difficulties in the calibration process. The same portable Preliminary analyses demonstrated that estimates of GR in- compressor and bendable chair were installed in the clinic at the cidence and progression did not differ between stages I (64 indi- Periodontal Department, and the examiners used the same medical viduals) and II (82 individuals). Thus, analyses were conducted headlight to conduct the examinations. At baseline, duplicate mea- dichotomously considering periodontitis stages I and II as the refer- sures were conducted in a total of 16 patients. During the fieldwork, ence category for comparisons with stages III or IV. 42 participants allowed to perform duplicate measurements. Before Prevalent cases of GR at baseline and follow-up were determined the follow-up examinations, duplicate measures were conducted in by the presence of at least 1 tooth with different thresholds of buc- 12 additional patients. cal and palatal/lingual GR. Prevalence estimates were also reported Baseline intra-examiner reliability for GR revealed weighted according to the absence and presence of periodontitis stages III/IV (±1 mm) Kappa values of 0.97 and 0.98, and the inter-examiner value to describe the pattern of GR changes at the individual level. was 0.84. During the baseline fieldwork, intra- and inter-examiner Data analyses were performed using a statistical package weighted Kappa values were 0.98, 0.99 and 0.91, respectively. (Stata 14 for Macintosh, STATA Corp., College Station, USA). Baseline intra-class correlation coefficient (ICC) values for means of Proportions, means, standard errors (SE) and 95% confidence in- GR ranged between 0.96 and 0.99. At the follow-up examination, tervals (95% CI) were calculated and reported. Pair-wise compari- weighted Kappa values for the four examiners ranged from 0.87 and sons of estimates were carried out using the Wald test considering 0.92, and ICCs varied from 0.95 to 0.98. the cluster of multiple sites per individual. The significance level was set at 5%. To account for differences between respondents and non-re- 2.7 | Statistical analyses spondents, the inverse probability weighting approach was applied for all commands in the analyses of the study (Hernán et al., 2004). Primary outcomes of this study were related to the incidence and Weights of the inverse probability of follow-up participation were progression of GR at buccal sites. Secondary outcomes included calculated for each individual and were applied to adjust all esti- estimates for GR at palatal/lingual sites. Teeth that were extracted mates. As a consequence, all estimates were reported with standard over the follow-up period were excluded from the analyses. errors or 95% CI instead of standard deviations. Two main outcomes were assessed in this study:

• Incident sites were defined as those that did not have GR at base- 3 | RESULTS line and developed GR after the follow-up period. This outcome was analysed as a dichotomous variable; Among the 1023 eligible individuals, 414 (40.5%) participated in the • The change in the position of the gingival margin quantified by follow-up evaluation (Figure 1). Baseline significant differences be- the subtraction of GR at follow-up from baseline. For this calcu- tween respondents (n = 414) and non-respondents (n = 609) were lation, only sites that had GR at the follow-up examination were observed for sex, age and socio-economic status (Table 1). No sig- included, representing progression of GR at sites that ended-up nificant differences were observed between respondents and non- with clinical recession after 4 years. This outcome was described respondents regarding education, smoking exposure, number of dichotomously using 1 mm, 2 mm and 3 mm as thresholds of change, as well as using the mean change overtime. teeth and gingival recession parameters. Among the 414 respondents, 7 individuals became edentulous Primary outcomes were reported according to periodontitis stages and 5 individuals had teeth that were unable to be examined. Thus, at the individual level and to interproximal CAL at the tooth level using the final sample of this study comprised 402 individuals with a total CAL data recorded in the follow-up examination. Teeth were divided of 6,862 teeth followed. The mean age of the sample at baseline was into those with interproximal CAL ≤1 mm (reference category) and 53.9 years, and 264 (65.7%) were women. At the follow-up examina- CAL ≥2 mm. Individuals were categorized into periodontitis stages tion, 256 (63.7%) had periodontitis stage III or IV. RIOS et al. | 119

TABLE 1 Baseline characteristics of Original sample Non-respondents Respondents the original sample, respondents and (N = 1023) (609) (n = 414) p* non-respondents Sex [n (%)] Male 635 (62.1) 255 (41.9) 143 (34.5) Female 398 (38.9) 354 (58.1) 271 (65.5) .02 Age [n (%)] 35-49 years 454 (44.4) 308 (50.6) 146 (35.3) 50-59 years 304 (29.7) 154 (25.3) 150 (36.2) ≥60 years 265 (25.9) 147 (24.1) 118 (28.5) <.001 Socio-economic status [n (%)] Low 505 (49.4) 324 (53.2) 181 (43.7) Medium 334 (32.7) 199 (32.7) 135 (32.6) High 184 (17.9) 86 (14.1) 98 (23.7) <.001 Educational status [n (%)] Low 198 (19.4) 114 (18.7) 84 (20.3) Medium 462 (45.2) 288 (47.3) 175 (42.3) High 362 (35.4) 207 (34.0) 155 (37.4) .28 Smoking [n (%)] Never-smoker 463 (45.3) 263 (43.2) 200 (48.3) Light-moderate 299 (29.2) 189 (31.0) 110 (26.6) Heavy 261 (25.5) 157 (25.8) 104 (25.1) .21 Number of teeth 18.8 ± 7.9 18.9 ± 7.9 18.6 ± 7.9 .51 [mean±SD] Gingival recession 0.66 ± 1.49 0.63 ± 1.51 0.71 ± 1.47 .41 [mean±SD] Gingival recession in ≥1 tooth [n (%)] ≥1 mm 1020 (99.7) 606 (99.5) 414 (100.0) .15 ≥3 mm 771 (75.4) 475 (78.0) 335 (80.9) .26 ≥5 mm 416 (40.7) 266 (43.7) 179 (43.2) .89

*Comparison between respondents and non-respondents; chi-square test.

The prevalence of individuals with buccal and palatal/lingual GR teeth with proximal CAL ≥2 mm at follow-up had incident buccal GR ≥1 mm was basically universal at baseline and follow-up, in individu- compared to 7.6% of teeth with proximal CAL ≤1 mm (p < .001). The als with periodontitis stages I/II and III/IV (Figure 2). The prevalence percentage of teeth with incident buccal GR was significantly higher of GR ≥3 mm in at least one tooth increased from 35.9% to 47.4% in the presence of proximal CAL ≥2 mm in individuals with (17.8%) (p < .001) at buccal sites and from 25.2 to 41.5% (p < .001) at palatal/ and without (20.2%) periodontitis stages III/IV compared to individ- lingual sites in individuals with periodontitis stages I/II. The prev- uals with CAL ≤1 mm and no periodontitis stages III/IV (4.2%). The alence of buccal and palatal/lingual GR ≥5 mm was small and did incidence of teeth with palatal/lingual GR was equal to 32.5% (95% not change significantly in these individuals. The prevalence of GR CI 29.7 - 35.3) (Figure 3B). At these sites, there was a significant dif- ≥3 mm and ≥5 mm was significantly higher in individuals with peri- ference between individuals with ≥20 teeth than those with <20 teeth odontitis stages III/IV. Prevalence rates also increased significantly and between those with and without periodontitis stages III/IV. No among these individuals. significant differences were observed between teeth that had or not At baseline, 3,356 (48.9%) teeth did not have GR at the buccal site. proximal CAL ≥2 mm. Individuals with periodontitis stages III/IV, with Among them, 1206 developed the condition overtime, resulting in an and without proximal CAL ≥2 mm, had higher percentage of incident overall incidence of buccal GR at the tooth level equal to 35.9% (95% palatal/lingual sites than those without periodontitis stages III/IV and CI 32.6 - 38.9) (Figure 3A). There was no significant difference in buc- with proximal CAL ≤1 mm. cal incident sites between individuals with ≥20 teeth than those with From all maxillary 1st molars, 64.3% developed buccal GR over- <20 teeth at baseline. Incident sites were significantly higher among time (Figure 4) comprising teeth with the higher incidence of buccal individuals with periodontitis stages III/IV (42.3%) compared to those GR among teeth with CAL ≤1 mm. They were followed by the 2nd with periodontitis stages I/II (29.5%) at the end of follow-up. 38.5% of maxillary premolar and lower central incisor. This pattern was similar 120 | RIOS et al.

FIGURE 2 Prevalence of individuals with buccal and palatal/lingual GR at baseline and follow-up, according to periodontitis stages

FIGURE 3 Percentage of incident sites with buccal (a) and palatal/lingual (b) gingival recession, according to number of teeth at baseline, presence of periodontitis stages III/IV at follow-up, presence of proximal CAL at follow-up, and the conjunction of presence of periodontitis and proximal CAL in the palatal/lingual sites. Among teeth with proximal CAL ≥2 mm, incisors, were the teeth with higher incidence of GR at buccal and mandibular 2nd premolars, maxillary 1st molars and mandibular cen- palatal/lingual sites. In individuals with periodontitis stages III/IV, tral incisors were teeth presenting higher incidence of buccal GR. posterior teeth presented higher incidence of GR. Mandibular central and lateral incisors were those with highest inci- Incidence of 1 mm, 2 mm and ≥3 mm was analysed selecting dence at palatal/lingual sites. In individuals with periodontitis stages 1,206 teeth that did not have buccal GR at baseline and ended the I/II, first and second molars, followed by the mandibular central follow-up with the condition (Table 2). In these teeth, the gingival RIOS et al. | 121

FIGURE 4 Intra-oral distribution of buccal and palatal/lingual sites that developed gingival recession overtime (% of teeth that did not have GR and presented it after the follow-up period), according to proximal CAL (a and b) and periodontitis stages III/IV (C and D) at follow- up 122 | RIOS et al.

TABLE 2 Percentage of buccal sites Change = 1 mm Change = 2 mm Change ≥3 mm with changes in the position of the mean ± SE p mean ± SE p mean ± SE p gingival margin (≥1 mm, ≥2 mm and ≥3 mm) ending the follow-up period Teeth that had no GR at baseline and ended with GR at follow-up (n = 1,206) with GR, according to number of teeth Overall 27.5 ± 1.6 43.7 ± 1.8 28.7 ± 1.8 at baseline, patients with or without Teeth at baseline periodontitis at follow-up, presence of proximal CAL at follow-up, and the <20 28.1 ± 3.0 Ref. 40.1 ± 3.2 Ref. 31.7 ± 3.8 Ref. conjunction of presence of periodontitis ≥20 27.4 ± 1.9 .84 44.9 ± 2.2 .22 27.7 ± 1.9 .34 and proximal CAL Periodontitis stages III/IV at follow-up No 29.9 ± 2.8 Ref. 46.9 ± 3.2 Ref. 23.1 ± 2.3 Ref. Yes 25.9 ± 1.9 .24 41.4 ± 2.1 .15 32.7 ± 2.4 <.001 Proximal CAL ≥2 mm at follow-up No 28.7 ± 8.9 Ref. 43.5 ± .9.4 Ref. 27.6 ± 10.2 Ref. Yes 27.6 ± 1.6 .89 43.7 ± 1.8 .99 28.7 ± 1.7 .92 Periodontitis and proximal CAL ≥2 mm No CAL/ 28.7 ± 16.9 Ref. 42.5 ± 7.3 Ref. 28.7 ± 12.5 Ref. No perio No CAL/ 28.8 ± 10.5 .69 44.1 ± 13.4 .92 27.1 ± 13.8 .93 Perio CAL/No 19.9 ± 2.9 .99 47.0 ± 3.2 .57 23.0 ± 2.3 .65 perio CAL/Perio 25.8 ± 1.9 .94 41.3 ± 2.0 .88 32.9 ± 2.4 .75 Teeth that had GR at both baseline and follow-up (n = 3,048) Overall 31.3 ± 1.1 8.7 ± 0.6 2.8 ± 0.4 Teeth at baseline <20 32.0 ± 1.5 Ref. 11.3 ± 1.0 Ref. 4.4 ± 0.8 Ref. ≥20 30.7 ± 1.5 0.53 7.0 ± 0.8 0.002 1.8 ± 0.4 .01 Periodontitis stages III/IV at follow-up No 25.3 ± 1.9 Ref. 2.5 ± 0.5 Ref. 0.5 ± 0.2 Ref. Yes 33.2 ± 1.3 0.001 10.8 ± 0.8 <0.001 3.6 ± 0.5 <.001 Proximal CAL ≥2 mm at follow-up No 42.0 ± 10.6 Ref. 0 ± 0 Ref. 0 ± 0 Ref. Yes 31.1 ± 1.0 0.31 8.8 ± 0.6 <0.001 2.9 ± 0.4 <.001 Periodontitis and proximal CAL ≥2 mm No CAL/ 31.5 ± 9.9 Ref. 0 ± 0 Ref. 0 ± 0 Ref. No perio No CAL/ 46.3 ± 14.2 0.39 0 ± 0 – 0 ± 0 – Perio CAL/No 25.2 ± 1.9 0.53 2.5 ± 0.6 <0.001 0.6 ± 0.3 .04 perio CAL/Perio 33.1 ± 1.2 0.87 10.8 ± 0.8 <0.001 3.6 ± 0.6 <.001

margin changed 1 mm in 27.5%, 2 mm in 43.7% and ≥3 mm in 28.7% with periodontitis stages I/II and III/IV, respectively; p < .001). Among of them. No significant differences were observed between catego- these progressive sites, 31.3%, 8.7% and 2.8% of these teeth demon- ries of number of teeth, periodontitis at the follow-up examination, strated progression of 1 mm, 2 mm and ≥3 mm, respectively. In these proximal CAL and the conjunction of periodontitis and proximal CAL. analyses, individuals with less teeth had significantly higher percent- Among 3,506 teeth that already had GR at baseline, 3,048 (86.9%) age of sites with progression of thresholds of 2 mm and ≥3 mm. Also, had progression of GR at buccal sites (90.9% and 76.3% in individuals individuals with periodontitis stages III/IV had significantly higher RIOS et al. | 123

TABLE 3 Mean changes (millimetres) in the position of the percentage of sites with progression equal to 2 mm and ≥3 mm than gingival margin at buccal and palatal/lingual sites, according to those with periodontitis stages I/II. Teeth without proximal CAL had number of teeth at baseline, presence of periodontitis at follow- less progression of 2 mm and ≥3 mm than those with proximal CAL. up, presence of proximal CAL at follow-up, and the conjunction of presence of periodontitis and proximal CAL The overall mean incidence of buccal GR was 2.11 mm (Table 3). Individuals with periodontitis stages III/IV had significantly greater INCIDENCE mean incidence (2.21 mm) than those with stages I/II (1.96 mm). No Teeth that had no GR PROGRESSION at baseline and ended Teeth that had GR other significant differences were observed in mean incidence. The with GR at follow-up at both baseline and overall mean progression of buccal GR was 0.40 mm. Individuals with (n = 1,206) follow-up (n = 3,048) <20 teeth had higher progression (0.51 mm) than those with ≥20 teeth (0.33 mm). Also, individuals with periodontitis stages III/IV had mean ± SE p mean ± SE p higher progression (0.49 mm) than those with stages I/II (0.14 mm). BUCCAL SITES In palatal/lingual sites, the overall mean incidence was 2.33 mm and Overall 2.11 ± 0.04 0.40 ± 0.03 progression was 0.48 mm. Individuals with periodontitis had signifi- Teeth at baseline cantly higher incidence and progression of GR at these sites. <20 2.25 ± 0.12 Ref. 0.51 ± 0.05 Ref. ≥20 2.05 ± 0.04 0.11 0.33 ± 0.04 .01 Periodontitis stages III/IV at follow-up 4 | DISCUSSION No 1.96 ± 0.05 Ref. 0.14 ± 0.05 Ref. Yes 2.21 ± 0.06 0.001 0.49 ± 0.04 <.001 This study showed that apical migration of the buccal and palatal/ Proximal CAL ≥2 mm at follow-up lingual gingival margin over an average period of 4 years occurred No 2.12 ± 0.25 Ref. 0.37 ± 0.13 Ref. in a significant proportion of individuals and teeth. In this regard, the overall prevalence of individuals with at least one tooth pre- Yes 2.10 ± 0.04 0.95 0.41 ± 0.03 .78 senting clinically relevant GR (≥3 mm) increased more than 10%. Periodontitis and proximal CAL ≥2 mm Comparatively with previous findings from the literature, these es- No CAL/ 2.00 ± 0.28 Ref. 0.31 ± 0.09 Ref. timates may be considered high. In a cohort of young adults from No perio New Zealand (Thomson et al., 2006), the prevalence of GR basically No CAL/ 2.18 ± 0.34 0.67 0.39 ± 0.17 .71 remained unchanged over 6 years. In the classic study of the natural Perio history of periodontal disease in man, 10% increase in the prevalence CAL/No 1.96 ± 0.05 0.89 0.13 ± 0.05 .09 of GR was observed only after each decade of life in the Norwegian perio cohort (Löe et al., 1992). These differences may be explained at least CAL/ 2.21 ± 0.06 0.46 0.49 ± 0.04 .09 in part by different oral hygiene habits, periodontal profiles, age of Perio the cohorts, possible genetic/ethnic determinants, time of follow-up PALATAL/LINGUAL SITES and exposure to risk factors. Overall 2.33 ± 0.04 0.48 ± 0.04 Importantly, the presence of periodontitis at the end of follow-up Teeth at baseline and proximal CAL had influenced GR changes overtime in this study. <20 2.48 ± 0.10 Ref. 0.63 ± 0.05 Ref. Thus, interpretation and comparisons with other studies should be ≥20 2.27 ± 0.04 0.05 0.32 ± 0.05 <.001 made considering similarities in regard to the presence of interprox- Periodontitis stages III/IV at follow-up imal CAL at patient and tooth levels. If we took individuals with periodontitis stages I/II for comparisons, they had GR progression No 2.17 ± 0.05 Ref. 0.23 ± 0.05 Ref. equal to 0.14 mm at teeth that already had GR at baseline. A some- Yes 2.44 ± 0.06 0.001 0.56 ± 0.05 <.001 what similar progression of 0.24 mm was observed in a study with Proximal CAL ≥2 mm at follow-up 40 dentists, but with a follow-up of 10 years (Matas et al., 2011). In a No 2.34 ± 0.14 Ref. 0.31 ± 0.12 Ref. study assessing GR only at Ramfjord teeth in 95 community-dwelling Yes 2.33 ± 0.04 0.99 0.48 ± 0.04 .15 adults followed for 10 years, progression of GR equalled 0.41 mm Periodontitis and proximal CAL ≥2 mm (Ship & Beck, 1996). Once again, differences in time of follow-up, No CAL/ 2.04 ± 0.22 Ref. 0.03 ± 0.15 Ref. periodontal examination protocols and source of the samples should No be considered for these comparisons. perio In this study, 76.3% of teeth that had buccal GR at baseline No CAL/ 2.54 ± 0.19 0.08 0.40 ± 0.14 .05 demonstrated progression of GR overtime in individuals with peri- Perio odontitis stages I/II. This estimate corroborates with a systematic CAL/No 2.18 ± 0.05 0.49 0.24 ± 0.06 .21 perio review which demonstrated that 78.1% of teeth experienced in- CAL/ 2.43 ± 0.06 0.07 0.56 ± 0.05 .001 crease in recession depth after the analyses of six studies evaluat- Perio ing progression of untreated GR in periodontally healthy individuals 124 | RIOS et al.

(Chambrone & Tatakis, 2016). Moreover, the overall proportion of One possible limitation of the present study was the rate of teeth presenting progression was much greater than that presenting attrition that was around 60%. However, this number is in accor- incidence of buccal GR (86.9% vs. 35.9%) in this study. This indicates dance with rates reached in other population-based follow-up that sites with GR are at higher risk for increasing recession depth studies (Baelum et al., 1997; Beck et al., 1997; Gilbert et al., 2005; than those without GR. Haas et al., 2012; Ismail et al., 1990; Machtei et al., 1999; Schätzle Nevertheless, the severity (magnitude) of change in buccal GR et al., 2003) and all estimates were generated applying a weight was much greater in incident teeth than progressive teeth. For ex- for non-response to minimize bias. Another methodological con- ample, the apical shift of the gingival margin was greater in sites that cern in this study was the reproducibility to assess GR after 4 years did not have GR at baseline (mean incidence =2.11 mm) compared to by different examiners. Due to time and budget constraints, the those that already had GR (mean progression =0.41 mm). Also, most two baseline examiners were unable to conduct the study alone at of the progressive sites had an increase of 1 mm in GR, whereas the the follow-up examination; consequently, two additional examin- majority of incident sites had changes of 2 mm or ≥3 mm. These ers participated. This may be considered a limitation in this study. findings were observed irrespectively of the periodontal status or However, the four examiners had great experience in conducting the presence of proximal CAL. The clinical implication of this finding epidemiological surveys and were carefully calibrated, as described is very important, in which preventive and therapeutic strategies in above. Assessment of reliability demonstrated very good coeffi- adults should also target sites without GR, and not only those with cients for intra- and inter-examiner ability to record GR. Also, to GR already established. Future research is needed to determine fac- overcome possible bias in the assessment of GR changes, different tors that may be related to differences in changes in recession depth thresholds of GR were analysed in this study. Thus, one may anal- between sites with and without GR. yse large thresholds changes of GR such as 2 mm and 3 mm, which The overall pattern of incidence and progression of GR was sim- are very unlike to be a result of examiner error, as demonstrated ilar in buccal and palatal/lingual sites. However, when analyses were by our calibration estimates, and may represent true changes in stratified for proximal CAL and periodontitis, differences became the gingival margin position. Other aspect that should be acknowl- evident. For instance, in individuals with periodontitis stages I/II and edged relates to the fact that the sample of this study is from an proximal CAL, the proportion of sites with incident GR was greater urban area and is exposed to dental treatment, and, consequently, in palatal/lingual (20.4%) than in buccal (4.7%) sites (Figure 3). changes in GR reported herein should be interpreted taking this Although it was not the aim of this study, it may be speculated that into consideration. this finding is probably due to greater plaque and calculus accu- In conclusion, the present study showed that incidence and mulation in palatal/lingual sites (Albandar & Kingman, 1999; Haas progression of GR occurred in a clinically relevant magnitude in a et al., 2019). Also, differences in mean changes of recession depth sample of adults from the general population of a Brazilian capital between categories of the conjunction variable of proximal CAL and city. Estimates varied according to the presence of proximal CAL and periodontitis were statistically significant for palatal/lingual sites periodontitis at the end of the follow-up. The following general con- and not for buccal sites (Table 3). This also corroborates the idea that clusions also can be made: incidence and progression of GR in palatal/lingual sites are probably related to periodontal disease rather than mechanical trauma, as it • the overall percentage of individuals with a clinically relevant de- has been also indicated in other studies (Albandar & Kingman, 1999; gree of GR (≥3 mm) increases approximately 12% at buccal and Löe et al., 1992). 16% at palatal/lingual sites; This is one of the very few cohort studies evaluating incidence • maxillary first molars and mandibular central incisors were the and progression of GR, prospectively, in an adult sample. The pop- teeth with higher incidence of GR at buccal sites; ulation-based design of the sample and the medium-term follow-up • more than 30% of buccal sites developed GR overtime; however, are also among the strengths of this study. Also, a variety of esti- this percentage was approximately 4% in teeth without proximal mates of incidence and progression of GR were reported to provide CAL from individuals with mild periodontitis; a broad range of the patterns of changes of the gingival margin. • the overall average proportion of sites that showed progression Additionally, estimates at buccal and palatal/lingual surfaces were was over 85%; and reported separately and according to different periodontal status • the overall mean change in the position of the gingival margin and interproximal CAL. was more than 2 mm in teeth without GR at baseline (incidence) Findings of the present study were reported according to peri- and lower than 0.5 mm in teeth that already had GR at baseline odontal status of individuals at the follow-up examination. Baseline (progression). data for CAL were not available; consequently, diagnosis of periodontitis was not possible at the beginning of the study. Also, it was Due to the negative impacts of GR in aesthetics, function and not possible to correlate incidence and progression of GR with the quality of life, strategies aiming to treat and prevent buccal and pal- development or progression of periodontitis overtime. This is still to atal/lingual GR should be encouraged in the individual and popula- be determined in future studies. tional levels. RIOS et al. | 125

CONFLICT OF INTEREST Hernán, M. A., Hernández-Díaz, S., & Robins, J. M. (2004). A structural approach to selection bias. Epidemiology, 15(5), 615-625. The authors declare no conflict of interest associated with the pre- Holtfreter, B., Schwahn, C., Biffar, R., & Kocher, T. (2009). Epidemiology sent study. of periodontal diseases in the Study of Health in Pomerania. Journal of Clinical Periodontology, 36(2), 114-123. https://doi. ORCID org/10.1111/j.1600-051X.2008.01361.x Ismail, A. I., Morrison, E. C., Burt, B. A., Caffesse, R. G., & Kavanagh, M. Alex N. Haas https://orcid.org/0000-0003-0531-6234 T. (1990). Natural history of periodontal disease in adults: findings from the Tecumseh Periodontal Disease Study, 1959–87. Journal of REFERENCES Dental Research, 69(2), 430-435. https://doi.org/10.1177/00220 ABEP (2013). Brazilian Association of Research Companies - Critério de clas- 345900690020201 sificação econômica Brasil (CCEB). Retrieved from http://www.abep. Jepsen, S., Caton, J. G., Albandar, J. M., Bissada, N. F., Bouchard, P., org/novo/Content.aspx?ContentID=835 Cortellini, P., Demirel, K., de Sanctis, M., Ercoli, C., Fan, J., Geurs, AAP (2001). Glossary of Periodontal Terms. In, 4th ed.. American Academy N. C., Hughes, F. J., Jin, L., Kantarci, A., Lalla, E., Madianos, P. of Periodontology. N., Matthews, D., McGuire, M. K., Mills, M. P., … Yamazaki, K. Albandar, J. M., & Kingman, A. (1999). Gingival recession, gingival bleed- (2018). Periodontal manifestations of systemic diseases and de- ing, and dental calculus in adults 30 years of age and older in the velopmental and acquired conditions: Consensus report of work- United States, 1988–1994. Journal of Periodontology, 70(1), 30-43. group 3 of the 2017 World Workshop on the Classification of https://doi.org/10.1902/jop.1999.70.1.30 Periodontal and Peri-Implant Diseases and Conditions. Journal Baelum, V., Luan, W. M., Chen, X., & Fejerskov, O. (1997). A 10-year of Clinical Periodontology, 45(Suppl 20), S219-S229. https://doi. study of the progression of destructive periodontal disease in adult org/10.1111/jcpe.12951 and elderly Chinese. Journal of Periodontology, 68(11), 1033-1042. Löe, H., Anerud, A., & Boysen, H. (1992). The natural history of peri- https://doi.org/10.1902/jop.1997.68.11.1033 odontal disease in man: prevalence, severity, and extent of gingi- Beck, J. D., Cusmano, L., Green-Helms, W., Koch, G. G., & Offenbacher, val recession. Journal of Periodontology, 63(6), 489-495. https://doi. S. (1997). A 5-year study of attachment loss in community-dwelling org/10.1902/jop.1992.63.6.489 older adults: incidence density. Journal of Periodontal Research, 32(6), Machtei, E. E., Hausmann, E., Dunford, R., Grossi, S., Ho, A., Davis, G., 506-515. Chandler, J., Zambon, J., & Genco, R. J. (1999). Longitudinal study of Chambrone, L., Salinas Ortega, M. A., Sukekava, F., Rotundo, R., predictive factors for periodontal disease and tooth loss. Journal of Kalemaj, Z., Buti, J., & Pini Prato, G. P. (2018). Root coverage pro- Clinical Periodontology, 26(6), 374-380. cedures for treating localised and multiple recession-type defects. Matas, F., Sentís, J., & Mendieta, C. (2011). Ten-year longitudinal study of Cochrane Database Systematic Review, 10, CD007161. https://doi. gingival recession in dentists. Journal of Clinical Periodontology, 38(12), org/10.1002/14651858.CD007161.pub3 1091-1098. https://doi.org/10.1111/j.1600-051X.2011.01799.x Chambrone, L., & Tatakis, D. N. (2016). Long-Term Outcomes of Rios, F. S., Costa, R. S., Moura, M. S., Jardim, J. J., Maltz, M., & Haas, Untreated Buccal Gingival Recessions: A Systematic Review and A. N. (2014). Estimates and multivariable risk assessment of gingi- Meta-Analysis. Journal of Periodontology, 87(7), 796-808. https:// val recession in the population of adults from Porto Alegre. Brazil. doi.org/10.1902/jop.2016.150625 J Clin Periodontol, 41(11), 1098-1107. https://doi.org/10.1111/ Costa, R. S., Rios, F. S., Moura, M. S., Jardim, J. J., Maltz, M., & Haas, A. N. jcpe.12303 (2014). Prevalence and risk indicators of dentin hypersensitivity in Sarfati, A., Bourgeois, D., Katsahian, S., Mora, F., & Bouchard, P. (2010). adult and elderly populations from Porto Alegre. Brazil. J Periodontol, Risk assessment for buccal gingival recession defects in an adult 85(9), 1247-1258. https://doi.org/10.1902/jop.2014.130728 population. Journal of Periodontology, 81(10), 1419-1425. https:// Dai, A., Huang, J. P., Ding, P. H., & Chen, L. L. (2019). Long-term stability doi.org/10.1902/jop.2010.100102 of root coverage procedures for single gingival recessions: A sys- Schätzle, M., Löe, H., Lang, N. P., Heitz-Mayfield, L. J., Bürgin, W., Anerud, tematic review and meta-analysis. Journal of Clinical Periodontology, A., & Boysen, H. (2003). Clinical course of chronic periodontitis. III. 46(5), 572-585. https://doi.org/10.1111/jcpe.13106 Patterns, variations and risks of attachment loss. Journal of Clinical Daprile, G., Gatto, M. R., & Checchi, L. (2007). The evolution of buccal Periodontology, 30(10), 909-918. gingival recessions in a student population: a 5-year follow-up. Serino, G., Wennström, J. L., Lindhe, J., & Eneroth, L. (1994). The preva- Journal of Periodontology, 78(4), 611-614. https://doi.org/10.1902/ lence and distribution of gingival recession in subjects with a high jop.2007.060277 standard of oral hygiene. Journal of Clinical Periodontology, 21(1), Gilbert, G. H., Shelton, B. J., & Fisher, M. A. (2005). Forty-eight-month 57-63. periodontal attachment loss incidence in a population-based co- Ship, J. A., & Beck, J. D. (1996). Ten-year longitudinal study of periodontal hort study: role of baseline status, incident tooth loss, and specific attachment loss in healthy adults. Oral Surgery, Oral Medicine, Oral behavioral factors. Journal of Periodontology, 76(7), 1161-1170. Pathology, Oral Radiology and Endodontics, 81(3), 281-290. https://doi.org/10.1902/jop.2005.76.7.1161 Stefanini, M., Marzadori, M., Aroca, S., Felice, P., Sangiorgi, M., & Haas, A. N., Gaio, E. J., Oppermann, R. V., Rösing, C. K., Albandar, Zucchelli, G. (2018). Decision making in root-coverage procedures J. M., & Susin, C. (2012). Pattern and rate of progression for the esthetic outcome. Periodontology 2000, 77(1), 54–64. of periodontal attachment loss in an urban population of https://doi.org/10.1111/prd.12205. South Brazil: a 5-years population-based prospective study. Susin, C., Haas, A. N., Oppermann, R. V., Haugejorden, O., & Albandar, J. Journal of Clinical Periodontology, 39(1), 1-9. https://doi. M. (2004). Gingival recession: epidemiology and risk indicators in a org/10.1111/j.1600-051X.2011.01818.x representative urban Brazilian population. Journal of Periodontology, Haas, A. N., Prado, R., Rios, F. S., Costa, R. D. S. A., Angst, P. D. M., Moura, 75(10), 1377-1386. https://doi.org/10.1902/jop.2004.75.10.1377 M. D. S., Maltz, M., & Jardim, J. J. (2019). Occurrence and predictors Thomson, W. M., Broadbent, J. M., Poulton, R., & Beck, J. D. (2006). of gingivitis and supragingival calculus in a population of Brazilian Changes in periodontal disease experience from 26 to 32 years adults. Braz Oral Res, 33, e036. https://doi.org/10.1590/1807- of age in a birth cohort. Journal of Periodontology, 77(6), 947-954. 3107bor-2019.vol33.0036 https://doi.org/10.1902/jop.2006.050319 126 | RIOS et al.

Tonetti, M. S., Greenwell, H., & Kornman, K. S. (2018). Staging and grading of periodontitis: Framework and proposal of a new classification How to cite this article: Rios FS, Costa RSA, Wagner TP, et al. and case definition. Journal of Clinical Periodontology, 45(Suppl 20), Incidence and progression of gingival recession over 4 years: S149-S161. https://doi.org/10.1111/jcpe.12945 A population-based longitudinal study. J Clin Periodontol. Wagner, T. P., Costa, R. S., Rios, F. S., Moura, M. S., Maltz, M., Jardim, J. J., & Haas, A. N. (2016). Gingival recession and oral health-related 2021;48:115–126. https://doi.org/10.1111/jcpe.13383 quality of life: a population-based cross-sectional study in Brazil. Community Dentistry and Oral Epidemiology, 44(4), 390-399. https:// doi.org/10.1111/cdoe.12226 Received: 26 June 2020 | Revised: 9 October 2020 | Accepted: 12 October 2020 DOI: 10.1111/jcpe.13390

S Y S T E M A T I C R E V I E W

P o s t - e x t r a c t i o n d i m e n s i o n a l c h a n g e s : A s y s t e m a t i c r e v i e w a n d meta-analysis

E m i l i o C o u s o - Q u e i r u g a1 | S a n d r a S t u h r1 | M u s t a f a T a t t a n1 | L e a n d r o C h a m b r o n e1,2,3 | G u s t a v o A v i l a - O r t i z1

1 Department of Periodontics , University of Iowa College of Dentistry , Iowa City , A b s t r a c t IA , USA Aim: To analyse the evidence pertaining to post-extraction dimensional changes in 2 Graduate Dentistry Program , Ibirapuera University , São Paulo , Brazil the alveolar ridge after unassisted socket healing. 3 Unit of Basic Oral Investigations , School Materials and Methods : The protocol of this PRISMA-compliant systematic review of Dentistry , Universidad El Bosque , (SRs) was registered in PROSPERO (CRD42020178857). A literature search to iden- Bogotá , Colombia tify studies that fulfilled the eligibility criteria was conducted. Data of interest were Correspondence extracted. Qualitative and random-effects meta-analyses were performed if at least Gustavo Avila-Ortiz, Department of Periodontics, University of Iowa College of two studies with comparable features and variables reported the same outcome of Dentistry, 801 Newton Road, Iowa City, IA interest. 52242, USA. E-mail: [email protected] Results : Twenty-eight articles were selected, of which 20 could be utilized for the conduction of quantitative analyses by method of assessment (i.e. clinical vs radiographic measurements) and location (i.e. non-molar vs molar sites). Pooled estimates revealed that mean horizontal, vertical mid-facial and mid-lingual ridge reduction assessed clinically in non-molar sites was 2.73 mm (95% CI: 2.36–3.11), 1.71 mm (95% CI: 1.30–2.12) and 1.44 mm (95% CI: 0.78–2.10), respectively. Mean horizontal, vertical mid-facial and mid-lingual ridge reduction assessed radiographically in non-molar sites was 2.54 mm (95% CI: 1.97–3.11), 1.65 mm (95% CI: 0.42–2.88) and 0.87 mm (95% CI: 0.36–1.38), respectively. Mean horizontal, vertical mid-facial and mid-lingual ridge reduction assessed radiographically in molar sites was 3.61 mm (95% CI: 3.24– 3.98), 1.46 mm (95% CI: 0.73–2.20) and 1.20 mm (95% CI: 0.56–1.83), respectively. C o n c l u s i o n: A variable amount of alveolar bone resorption occurs after unassisted socket healing depending on tooth type.

K E Y W O R D S alveolar bone loss , bone resorption, tooth extraction

1 | INTRODUCTION maintained in a status compatible with adequate aesthetics, function and/or health, or for strategic reasons (Kao, 2008 ; Tonetti The main goal of dental therapy is to enhance and maintain the et al., 2000). Beyond its potential impact in quality of life, tooth general health and well-being of patients (Chapple & Wilson, extraction causes a local physiologic disruption that results in an 2014 ). Dental extraction is indicated when teeth cannot be initial inflammatory response and, subsequently, a variable degree

J Clin Periodontol. 2021;48:127–145. wileyonlinelibrary.com/journal/jcpe | 127 128 | COUSO-QUEIRUGA ET AL. of alveolar ridge atrophy, which is primarily related to bone resorption (Araujo & Lindhe, 2005 ; Cardaropoli et al., 2003; Evian et al., Clinical Relevance 1982; Trombelli et al., 2008 ). Numerous pre-clinical and clinical Scientific rationale : There is a lack of evidence from previous studies have reported that most of the bone remodelling occurs systematic reviews (SRs) regarding post-extraction dimen- within the first few weeks after tooth extraction and it is more sional changes in the alveolar ridge after unassisted socket accentuated on the facio-coronal aspect of the ridge (Araujo & healing in function of local, systemic and surgical factors. Lindhe, 2005 ; Avila-Ortiz et al., 2020 ; Chappuis et al., 2013 , 2015; Principal findings: Horizontal and vertical alveolar ridge re- Discepoli et al., 2013 ). duction was consistently reported in all articles included Predicting the extent and magnitude of post-extraction dimen- in this SR. These dimensional alterations are more accen- sional changes in function of specific patient and site characteris- tuated on the coronal aspect of the ridge and can be pri- tics is particularly important to make judicious clinical decisions marily attributed to bone resorption. Increased horizontal when tooth replacement therapy is planned and alveolar ridge ridge reduction was observed in molar sites compared with preservation, with or without immediate implant placement, is non-molar sites. Facial bone thickness upon extraction was being considered (Avila-Ortiz et al., 2019 ; Clementini et al., 2019). strongly associated with the extent and magnitude of al- Clinical evidence indicates that some phenotypic factors may influ- veolar bone resorption, specifically in non-molar sites. ence bone resorption patterns (Avila-Ortiz et al., 2020 ; Chappuis Practical implications: Alveolar bone resorption is a physiologic et al., 2013). Two previous systematic reviews (SRs) on this topic phenomenon that occurs as a consequence of tooth extrac- evaluated bone and soft tissue changes (Tan et al., 2012 ; Van der tion. Dimensional changes seem to be larger in molar sites, Weijden et al., 2009 ). Studies included in these SRs were markedly particularly in the horizontal dimension. However, this was not heterogeneous, and no attempt to explore the role of different directly correlated with the need for ancillary bone grafting variables on the outcomes was made. The aim of this SR was to prior to or at the time of implant placement. This information answer the following focused question: What are the dimensional should be taken into consideration when making clinical deci- changes in the alveolar ridge that result from unassisted socket sions pertaining to the management of fresh extraction sites. healing in adult human subjects in function of relevant local, systemic and surgical factors?

2 | MATERIALS AND METHODS A Dimensional changes in the alveolar ridge:

The protocol of this review was previously registered in the A.1. Bone linear measurements [horizontal, mid-facial, mid-lingual] International Prospective Register of Systematic Reviews in mm assessed clinically upon surgical re-entry or radio- (PROSPERO) with the identification code CRD42020178857. graphically (primary outcome of interest). This review fully adhered to the guidelines of the Preferred A.2. Bone volumetric measurements in mm3 assessed radiographi- Reporting Items of Systematic Reviews and Meta-Analyses cally using DICOM files. (Moher, Liberati, Tetzlaff, Altman, & Group) statement (Moher A.3. Soft or hard tissue linear [horizontal, mid-facial, mid-lingual] or et al., 2009 ). volumetric assessments, in mm and mm3 , respectively, assessed digitally using stereolithography (STL) files.

2 . 1 | PICO Question A Incidence and type of complications during the healing period. “What are the dimensional changes of the alveolar ridge that result B Feasibility of implant placement / need for ancillary site de- from unassisted socket healing in adult human subjects in function velopment procedures prior to or simultaneous with implant of relevant local, systemic and surgical factors?” placement. Population C Patient-reported outcome measures (PROMs): discomfort, per- Adult human subjects (>18 years of age) requiring single tooth ceived benefit and quality-of-life changes. extraction, except mandibular incisors and third molars. Intervention In addition, the effect of several factors on the outcomes of inter- The intervention of interest was standard-of-care tooth ex- est was explored. These factors were divided into four categories: (1) traction with or without socket curettage, irrigation and suturing as patient-related (i.e. smoking habits, diabetes, history of periodontitis), required. (2) local anatomical (i.e. non-molar or molar site, socket integrity), (3) Comparison phenotypic (keratinized tissue width, soft tissue and bone plate thick- No comparison groups. ness, supracrestal tissue attachment dimension) and (4) surgical-re- Outcomes lated (i.e. flap elevation, primary or secondary intention healing). COUSO-QUEIRUGA ET AL. | 129

2 . 2 | Criteria for study inclusion reason(s) for extraction, socket anatomy (i.e. non-molar or molar site), supracrestal tissue attachment (previously termed biologic Proper RCTs, non-RCTs and prospective case series published in width), keratinized tissue width, facial and lingual soft tissue thick- English language that reported the outcomes of unassisted socket ness, facial and lingual bone plate thickness, flap elevation, primary healing after tooth extraction were included. closure and total healing period. Any missing data that could contrib- Clinical scenarios included in this review were (1) intact or ute to the conduction of this review were requested from the cor- well-preserved extraction sites and (2) partially damaged sites responding authors of the original articles via email communication. presenting a minimum of 50% of alveolar bone height loss on any If no response was received, the requested data set was excluded wall. Studies involving any additional intervention that may have from the final analysis. influenced the outcomes of interest (i.e. collagen plug or autologous blood-derived product placement in the socket or delivery of an immediate removable mucosa-supported prosthesis) were 2 . 6 | Risk-of-bias assessment excluded. For inclusion, studies must have recruited adult human subjects (>18 years of age) requiring single tooth extraction, except The risk-of-bias analyses of each RCT included in the final selection mandibular incisors and third molars. A minimum post-operative fol- were independently performed by two authors (M.T. and S.S.) using low-up period of 2 months (≥8 weeks) and a maximum of 12 months the Cochrane Collaboration' s tool for assessing risk of bias in rand- (≤52 weeks) were required for inclusion. No minimum number of pa- omized clinical trials (Higgins et al., 2019 ). RCTs were categorized tients per study was necessary for inclusion. Finally, studies must as being at low, some concerns or high risk of bias. Non-RCTs were have reported at least one of the outcomes of interest aforemen- assessed using the ROBINS-I tool (Sterne et al., 2016 ). Each non-RCT tioned to be eligible. was categorized as having no information or being at low, moderate, serious or critical risk of bias (see Appendix 1 for more information). Disagreement between reviewers was resolved by open discussion. 2 . 3 | Search methods In case that no agreement could be achieved, the final decision was made by another co-author (E.C.Q.). Three electronic databases were searched, namely National Library of Medicine (MEDLINE–PubMed), Cochrane Central Register of Controlled Trials (CENTRAL) and EMBASE using a specific MEDLINE 2 . 7 | Data synthesis strategy. The last electronic search was conducted on 15 August 2020. Additionally, a thorough hand search was performed (see Data were organized into evidence tables, and a descriptive sum- Appendix 1 for more information). mary was performed to determine the quantity of data and study variations (i.e. study subjects, extraction site features and results). Following article selection, Cohen' s kappa coefficient (k ) was per- 2 . 4 | Article selection formed to assess inter-examiner agreement. Continuous data were pooled into random-effects meta-analyses and expressed as initial Two reviewers (E.C.Q. and M.T.) independently read the title and and final averages to calculate mean differences (MD) with their as- abstract of the entries obtained from the literature searches and sociated 95% confidence intervals (CI). The analyses were performed made a preliminary selection. Then, both reviewers read individually using the generic inverse variance statistical method where the MD through the full-text versions of the potentially eligible studies. Final and standard error (SE) are entered for all studies (Stedman et al., article selection based on the aforementioned eligibility criteria was 2011 ). Statistical heterogeneity was assessed by calculation of the performed. Both reviewers had an open discussion when disagree- Q statistic. The significance of discrepancies in the estimates of the ment regarding the final selection of an article occurred. If no agree- treatment effects from the different trials was assessed by means of ment was achieved, another co-author (S.S.) made the final decision. Cochrane' s test for heterogeneity and the I 2 statistic. Analyses were performed using RevMan software (Review Manager 5, version 5.3; Nordic Cochrane Centre). 2 . 5 | Data extraction

Data extraction was done separately by two of the authors (E.C.Q 3 | RESULTS and M.T.). The accuracy of the data was verified independently by a third author (S.S.). In addition to the outcomes of interest and fac- 3 . 1 | Search results tors previously outlined, additional data extracted included year of publication and first author, study design, initial number of partici- The initial database search yielded a total of 11,075 entries, of pants and distribution by groups, number of drop-outs, age and gen- which 5180 were found in PubMed, 1960 in EMBASE and 3935 in der distribution, smoking habits, diabetes, history of periodontitis, CENTRAL. Following duplicate removal, 7356 entries remained. 130 | COUSO-QUEIRUGA ET AL.

After title and abstract screening, a total of 106 articles were se- 3 . 2 | Characteristics of the included studies lected for full-text review. Four additional articles were identified through hand searching. Seventy-eight of these articles were The general characteristics of the 28 selected studies including excluded after full-text review, the reasons for which are sum- study design, population, setting, extraction site features and fol- marized in Table S1 . Thus, the final selection was comprised of low-up (healing) time are displayed in Table S2 . 28 articles, of which 25 were RCTs (Aimetti et al., 2009 ; Araujo et al., 2015; Avila-Ortiz et al., 2020 ; Barone et al., 2008 , 2017; Canellas et al., 2020 ; Cha et al., 2019 ; Clementini et al., 2019; 3 . 3 | Risk of bias in selected studies Festa et al., 2013 ; Fiorellini et al., 2005 ; Hauser et al., 2013 ; Iasella et al., 2003; Jung et al., 2013 ; Karaca et al., 2015 ; Lim Fourteen RCTs were categorized as presenting low risk (Araujo et al., 2019; Machtei et al., 2019 ; Pang et al., 2016 ; Pelegrine et al., 2015; Avila-Ortiz et al., 2020 ; Barone et al., 2008 , 2017; et al., 2010; Qabbani et al., 2017 ; Rasperini et al., 2010 ; Sisti et al., Canellas et al., 2020 ; Cha et al., 2019; Clementini et al., 2019; Iasella 2012; Spinato et al., 2014 ; Sun et al., 2019 ; Thalmair et al., 2013 ; et al., 2003; Jung et al., 2013 ; Lim et al., 2019 ; Machtei et al., 2019 ; Thoma et al., 2020 ) and three were non-RCTs (Jiang et al., 2017 ; Rasperini et al., 2010 ; Sisti et al., 2012 ; Sun et al., 2019 ), two RCTs as Sbordone et al., 2017 ; Zhao et al., 2018 ). Kappa scores for inter- some concerns (Hauser et al., 2013 ; Pang et al., 2016 ) and nine RCTs examiner agreement for title and abstract review as well as full- as high risk of bias (Aimetti et al., 2009 ; Festa et al., 2013 ; Fiorellini text review were 0.91 and 0.82, respectively. The article selection et al., 2005; Karaca et al., 2015 ; Pelegrine et al., 2010 ; Qabbani et al., process is depicted in Figure 1 . 2017; Spinato et al., 2014 ; Thalmair et al., 2013 ; Thoma et al., 2020 ).

FIGURE 1 Search and article selection process COUSO-QUEIRUGA ET AL. | 131

Two non-RCTs were categorized as low risk (Sbordone et al., 2017 ; Zhao et al., 2018 ), and the remaining one was categorized as serious risk (Jiang et al., 2017 ) as displayed in Figures 2 and 3.

3 . 4 | Qualitative assessment of outcomes

The extracted data pertaining to the clinical and radiographic outcomes of interest are displayed in Table 1 , while data on STL analyses and patient-reported outcomes are shown in Table S3 .

3 . 4 . 1 | Dimensional outcomes

Clinical horizontal bone changes Differences in horizontal (facio-lingual) bone changes between baseline and final follow-up assessed by direct clinical measurements were reported in ten studies. Nine studies included only non- molar teeth (Aimetti et al., 2009 ; Barone et al., 2008 , 2017; Festa et al., 2013 ; Hauser et al., 2013 ; Iasella et al., 2003 ; Machtei et al., 2019; Pelegrine et al., 2010 ; Spinato et al., 2014 ), whereas one study included a mix of non-molars and molar teeth, showing higher net amounts of horizontal alveolar ridge resorption in non-molars vs molars (−3.60 ± 0.72 vs −3.13 ± 0.35 mm) (Barone et al., 2017 ). In FIGURE 3 Risk-of-bias assessment of non-RCTs six studies, a stent or template was used (Barone et al., 2008 ; Festa et al., 2013 ; Hauser et al., 2013 ; Machtei et al., 2019 ; Pelegrine et al., 2010; Spinato et al., 2014 ). In the trial by Spinato et al. (2014 ), patients were divided into subgroups according to the thickness of in nine studies, with all reporting differences in mid-facial vertical the facial bone plate in non-molar maxillary sites. Sites presenting bone changes (Aimetti et al., 2009 ; Barone et al., 2008 , 2017; Festa a thick facial bone plate (>1 mm) showed less horizontal resorption et al., 2013 ; Iasella et al., 2003 ; Machtei et al., 2019 ; Pelegrine than those with a thin plate (≤1 mm) (1.17 ± 0.41 vs 2.67 ± 0.52 mm). et al., 2010; Rasperini et al., 2010 ; Spinato et al., 2014 ). Six trials reported differences in mid-lingual bone changes (Aimetti et al., Clinical vertical bone changes 2009; Barone et al., 2008 , 2017 ; Festa et al., 2013 ; Iasella et al., Differences in vertical bone changes between baseline and final 2003; Spinato et al., 2014 ). Two studies did not utilize a stent to follow-up assessed by direct clinical measurements were reported assess the changes (Pelegrine et al., 2010 ; Spinato et al., 2014 ). In

FIGURE 2 Risk-of-bias assessment of RCTs 132 | COUSO-QUEIRUGA ET AL.

3 (Continues) N/R V o l u m echange in t mm r i c N/R (mean ±SD) ±SD) (mean N/R N/R N/R N/R V e r t i clingual change a l m i (mean d mm in - ±SD) N/R N/R N/R N/R (only changes reported) 1.17 ± 1.23 1.23 ± 1.17 N/R V e r t i cfacial change a in l m i ±SD) d (mean mm - N/R N/R − N/R changes reported) N/R Radiographic outcomes (CBCT) Horizontal change in mm ±SD) (mean N/R +0.57 ± 2.56 (only N/R N/R

number of events events of number of oral oedema and erythema in the test groups compared with the control, but details not specified. None None Incidence and type of complications None None Oral oedema (75%) Mouth pain (68%) Oral erythema (46%) There were a greater N/R None N/R Feasibility of placement implant with no additional grafting bone N/R 9 of 20 sites (45%) N/R N/R .7 ± 1.7 ± 1.7.7 changes changes reported) changes reported) 8.8 ± 0.9 9 0.4 ± 1.0 (only 3.0 ± 1.6 (only N/R V e r t i clingual bone a l m i dchange in mm - ±SD) (mean − N/R − Baseline: 3 months: 2.46 ± 0.4 1.29 ± 0.4 changes changes reported) changes changes reported) ± 2.211.1 0.9 ± 1.6 (only 3.6 ± 1.5 (only Baseline: 6 months: Vertical Vertical mid-facial bone change in mm ±SD) (mean − N/R − Baseline: ± 2 9.9 3 months: Baseline: ± 0.49 7.38 6 months: 4.92 ± 0.86 Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm Baseline: ± 1.0 9.1 4 to 6 months: 6.4 ± 2.2 N/R Baseline: ± 0.7 10 3 months: 6.8 ± 1.3 Baseline: 10.8 ± 0.8 months:7–9 8.1 ± 1.4 Q

uantitative data of included studies pertaining to the clinical and radiographic outcomes of interest ) ) ) ) ) 2010 2010 2003 2005 2008 2009 et al. ( ( ( ( ( Pelegrine Publication(s) Publication(s) Author and (s) Year Iasella et al. Fiorellini et al. Barone et al. Aimetti et al. TABLE TABLE 1 | 133

3 (Continues) N/R (mean ±SD) ±SD) (mean V o l u m echange in t mm r i c N/R N/R N/R N/R changes changes reported) 0.6 ± 0.6 (only − V e r t i clingual change a l m i (mean d mm in - ±SD) N/R N/R N/R N/R changes changes reported) 0.5 ± 0.9 (only − V e r t i cfacial change a in l m i ±SD) d (mean mm - N/R N/R N/R N/R 0.8 ± 1.2 (only 3.3 ± 2.0 1.7 ± 0.8 − − − changes reported) At 1 mm: At 3 mm: At 5 mm: Radiographic outcomes (CBCT) mm in change Horizontal ±SD) (mean N/R N/R N/R N/R None None Incidence and type of complications None None N/R None None None None

(100%) defects: 2 of 3 (66%) sites: 3 of 5 (60%) N/R Feasibility of placement implant with no additional grafting bone 0 of 20 sites No significant Partially damaged 7 of 7 (100%) N/R changes changes reported) 2.4 ± 1.6 (only N/R V e r t i clingual bone a l m i dchange in mm - ±SD) (mean N/R − N/R N/R 5.7 ± 4.2 defects/ all walls remaining: − changes changes reported) sites: sites: (no reported final data) 3.1 ± 1.3 (only No significant Initial: 8.3 ± 3.2 missing Final: Changes: N/R Vertical Vertical mid-facial bone change in mm ±SD) (mean N/R − Partially damaged Initial: 16.4 ± 4 missing Final: Changes: 0.6 ± 3.9 N/R tangent of the dental mid-point the at arch of the extraction site approximately 4 mm under the level of the gingiva of the adjacent teeth and corresponded to the distance between the most prominent points buccally and orally) (only changes reported) 0.43 (perpendicular to the N/R N/R Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm N/R Baseline: ± 1 9.9 6 months: 6.2 ± 1.3 − Con tinued ) ) ) ) ) 2010 2010 2013 2013 2012 2012 2013 2013 2013 et al. ( ( ( ( ( ung et al. Publication(s) Publication(s) Author and (s) Year Rasperini Rasperini Sisti et al. Festa et al. Hauser et al. J TABLE 1 134 | COUSO-QUEIRUGA ET AL.

3 ontinues) (C N/R (mean ±SD) ±SD) (mean V o l u m echange in t mm r i c N/R N/R N/R N/R N/R N/R

(median values) 9.3 ± 2.69.3 ± 2.4 7.9 N/R V e r t i clingual change a l m i (mean d mm in - ±SD) Baseline: 6.96 3 months: 5.84 N/R Baseline: 4 months: N/R N/R N/R

) ) s (median value 9.4 ± 1.6 9.4 5.8 ± 2.2 N/R V e r t i cfacial change a in l m i ±SD) d (mean mm -

Baseline: 7.26 3 months: 5.98 N/R Baseline: 4 months: N/R N/R N/R 3.26 ± 0.44 3.82 ± 0.33 3.82 ± 0.33 − − N/R

Radiographic outcomes (CBCT) mm in change Horizontal ±SD) (mean No significant defects: significant No Partially damaged sites: N/R N/R N/R N/R

Baseline: Baseline: 6.97 ± 0.91 4 months: 3.85 ± 1.14 None Incidence and type of complications N/R None None None None None N/R Feasibility of placement implant with no additional grafting bone N/R N/R N/R N/R N/R

0/12 (100%)

1.0 ± 0.63 − 0.5 ± 0.55 − (only changes reported) N/R V e r t i clingual bone a l m i dchange in mm - ±SD) (mean N/R Thin: N/R N/R Thick: Thick: N/R N/R

0.50 ± 0.55 0.55 ± 0.50 1.17 ± 0.41.17 − − (only changes reported) N/R Vertical Vertical mid-facial bone change in mm ±SD) (mean Thin: N/R Thick: Thick: N/R N/R N/R N/R 1.17 ± 0.41 (only1.17 2.67 ± 0.52 − − changes reported) Thin: N/R Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm Thick: Thick: N/R N/R N/R N/R N/R Continued ) ) ) ) ) ) ) 2014 2014 2013 2013 2016 2016 2015 2015 2015 2015 2017 2017 2017 2017 ( ( ( ( ( et al. ( ( Publication(s) Publication(s) Author and (s) Year Spinato et al. Thalmair et al. al. et Thalmair Pang et al. Karaca et al. Araújo et al. Sbordone Sbordone Jiang et al. TABLE 1 COUSO-QUEIRUGA ET AL. | 135

3 ontinues) (C (mean ±SD) ±SD) (mean V o l u m echange in t mm r i c N/R N/R N/R N/R

2.16 ± 1.62 height height >lingual ridge height: - 0.55 ± 1.00 height lingual ridge height: − hei V e r t i cfacial change a in l m i ±SD) d (mean mm - N/R N/R N/R Buccal ridge Buccal ridge 8.22 ± 1.30 9.22 ± 1.39 10.40 ± 1.38 6.51 ± 0.79 8.31 ± 1.79 ± 2.30 9.76

Radiographic outcomes (CBCT) mm in change Horizontal ±SD) (mean Baseline: Baseline: At 0 mm: ± 1.11 6.74 At 3 mm apical: At 5 mm apical: At 7 mm apical: 9 months: At 0 mm: 4.90 ± 1.13 At 3 mm apical: At 5 mm apical: At 7 mm apical: N/R N/R N/R Incidence and type of complications None N/R N/R None Feasibility of placement implant with no additional grafting bone N/R N/R N/R N/R

2.03 ± 0.72 2.13 ± 0.84 − − (only changes reported) V e r t i clingual bone a l m i dchange in mm - ±SD) (mean N/R N/R Anterior: Posterior: N/R

2.10 ± 0.662.10 2.25 ± 0.46 9.88 ± 3.6 9.88 ± 3.6 − − reported) Vertical Vertical mid-facial bone change in mm ±SD) (mean N/R Baseline: ± 4.2 7.69 4.5 ± 0.4 months: Anterior: Posterior: (only changes N/R 3.60 ± 0.72 − Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm N/R Baseline: At 3 mm: 8.31 ± 1.4 At 6 mm: 8.30 ± 1.9 4.5 ± 0.4 months: At 3 mm: 5.35 ± 1.2 At 6 mm: 6.50 ± 1.7 Anterior: Posterior: ± 0.35 - 3.13 (only changes reported) N/R Continued ) ) ) ) 2017 2017 2017 2017 2019 2019 2018 2018 ( ( ( ( Publication(s) Publication(s) Author and (s) Year Qabbani et al. Barone et al. Machtei et al. Zhao et al. TABLE 1 136 | COUSO-QUEIRUGA ET AL.

3 (Continues) N/R N/R (mean ±SD) ±SD) (mean V o l u m echange in t mm r i c N/R N/R

0.9 ± 0.8 1.3 ± 1.8 (only changes reported) (only changes reported) − (only changes reported) − 0.21 ± 0.31 0.31 ± 0.21 1.20 ± 0.96 N/R − V e r t i clingual change a l m i (mean d mm in - ±SD) Molar sites: − Non-molar sites:

3 ± 2.6 1.9 ± 2.4 (only changes reported) − (only changes reported) − (only changes reported) 0.83 ± 1.14 1.33 ± 1.11 1.11 ± 1.33 N/R − V e r t i cfacial change a in l m i ±SD) d (mean mm - Non-molar sites: − Molar sites:

0.84 ± 0.75 0.75 0.84 ± 4.44 ± 3.71 6.5 ± 4.2 3.9 ± 3.2 3.37 ± 1.55 2.41 ± 1.97 2.9 ± 3.7 2.27 ± 1.15 1.8 ± 1.1 1 1.88 ± 1.55 1.55 1.88 ± 1.0 ± 0.6 (only

− − − − − − .0 ± 0.9 − − − − − − 9.69) 12.64) 15.40) 6.58) 11.03) 14.13) changes reported) (only changes reported) Non-molar sites: At 1 mm: At 3 mm: At 5 mm: At 1 mm: At 3 mm: At 5 mm: (only changes reported) Baseline: Baseline: CI At 2 mm: (5.79, 7.81 At 4 mm: 10.77 CI (8.85, At 6 mm: 14.09 (12.44, 6 months: At 2 mm: 4.27 CI (2.17, At 4 mm: 8.97 CI (7.05, At 6 mm: (11.38, 12.74 Radiographic outcomes (CBCT) mm in change Horizontal ±SD) (mean Molar sites: At 1 mm: At 3 mm: At 5 mm: At 1 mm: At 3 mm: At 5 mm: None None None None None None Incidence and type of complications None None

needed sinus sinus needed augmentation procedure 4 of 13 (31%)4 of 13 N/R 0 of 14 (100%) 0 of 14 All of them Feasibility of placement implant with no additional grafting bone 5 of 8 (62.5%) N/R N/R N/R V e r t i clingual bone a l m i dchange in mm - ±SD) (mean N/R N/R N/R N/R Vertical Vertical mid-facial bone change in mm ±SD) (mean N/R N/R N/R N/R Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm N/R Continued ) ) ) ) 2019 2019 2019 2019 2019 2019 2019 2019 ( et al. ( ( ( Publication(s) Publication(s) Author and (s) Year Sun et al. Clementini Clementini Cha et al. Lim et al. TABLE 1 COUSO-QUEIRUGA ET AL. | 137

3 (only changes reported) 44.87 ± 200.20 154.51 ± 69.35 N/R − − V o l u m echange in t mm r i c (mean ±SD) ±SD) (mean 0.91 ± 0.63 0.63 ± 0.91 1.24 ± 1.15 1.15 ± 1.24 N/R − − V e r t i clingual change a l m i (mean d mm in - ±SD) (only changes reported) 1.39 ± 1.28 1.28 ± 1.39 1.7 ± 1.60 N/R − − V e r t i cfacial change a in l m i ±SD) d (mean mm -

1.26) 1.26)

− 2.27 ± 1.21 1.67 ± 1.10 1.10 ± 1.67 1.04 ± 1.08 − − − 2.10 to to 2.10 − (only changes reported) 1.67 ± 1.17 1.17 ± 1.67 At 1 mm: At 3 mm: At 5 mm: − (CI: N/R Radiographic outcomes (CBCT) Horizontal change in mm ±SD) (mean to be removed 2 months after placement due to lack of osseointegration. (not specified which group) One implant had None None None None Incidence and type of complications

= 1); = 5); = 1) n n n <1.5 mm ( Dehiscence type defect after implant placement ( intra-bony type defect after implant 1 placement ( N/R 14 of 27 (51.9%) of 27 (51.9%) 14 1 of 11 (9.1%) (9.1%) 1 of 11 Thin buccal bone Feasibility of placement implant with no additional grafting bone N/R N/R N/R V e r t i clingual bone a l m i dchange in mm - ±SD) (mean N/R N/R N/R Vertical Vertical mid-facial bone change in mm ±SD) (mean N/R N/R N/R Clinical Clinical outcomes Horizontal bone change in ±SD) (mean mm Continued ) ) ) 2020 2020 2020 ( ( ( Canellas et al. al. Canellas et Avila et al. Publication(s) Publication(s) Author and (s) Year Thoma et al. TABLE 1 138 | COUSO-QUEIRUGA ET AL. one of them, vertical changes were assessed using titanium screws Linear and volumetric changes assessed in STL files as reference points, which were inserted into the facial and lingual Linear changes in alveolar ridge soft tissue contour assessed in STL bone plate (Pelegrine et al., 2010 ). In the other trial, a periodontal files were reported in two studies (Thalmair et al., 2013 ; Thoma et al., probe using the CEJs of the adjacent teeth as a horizontal refer- 2020 ). Average ridge contour reduction was higher in the study con- ence line was utilized (Spinato et al., 2014 ). For all vertical bone ducted by Thalmair et al. compared with the one by Thoma and col- changes, non-molar sites consistently rendered more mid-facial laborators (−2.29 ± 1.1 vs −1.8 ± 0.8 mm). Thalmair and co-workers than mid-lingual bone resorption. Interestingly, the trial by Spinato assessed the measurements at the most coronal contour line of the et al. ( 2014 ) showed exactly the same amount of mid-facial and alveolar ridge. Meanwhile, Thoma et al. assessed horizontal linear mid-lingual resorption in non-molar maxillary sites when a thick fa- changes at 3 and 5 mm apical from the mucosal margin. Only three cial bone plate (>1 mm) was present (−0.5 ± 0.55 vs −0.5 ± 0.55 mm, studies assessed soft tissue volumetric changes between baseline respectively). In one study (Rasperini et al., 2010 ), more bone re- and final follow-up (Avila-Ortiz et al., 2020 ; Sbordone et al., 2017 ; modelling was observed when the molar socket was intact vs a Thalmair et al., 2013). Significant value discrepancies were observed molar socket that was severely damaged. Barone and co-workers in non-molar teeth (initial: 990 ± 0.40 mm3 and final: 540 ± 0.20 mm3 ) (Barone et al., 2017 ), however, reported similar mid-facial and mid- in the study by Sbordone et al. compared with the study by Thalmair lingual bone changes in the posterior and anterior regions (mid- and collaborators (−41.41 ± 15.96 mm3 ), which reported lower mean facial: −2.25 ± 0.46 vs −2.10 ± 0.66 mm / mid-lingual: −2.13 ± 0.84, values than Avila-Ortiz and colleagues (−107.27 ± 48.11 mm3 ). Data −2.03 ± 0.72 mm, respectively), although the distribution of socket on bone tissue volumetric changes assessed by stereolithography integrity was not reported. were not reported in any of the selected studies.

Radiographic horizontal bone changes Differences in horizontal bone changes between baseline and final 3 . 4 . 2 | Complications follow-up assessed in CBCT scan images were reported in eleven studies (Avila-Ortiz et al., 2020 ; Canellas et al., 2020 ; Cha et al., Only one study reported the occurrence of complications (Fiorellini 2019; Clementini et al., 2019; Fiorellini et al., 2005 ; Jiang et al., 2017 ; et al., 2005), twenty-two studies reported uneventful healing Jung et al., 2013 ; Lim et al., 2019 ; Pang et al., 2016 ; Qabbani et al., (Aimetti et al., 2009; Araujo et al., 2015 ; Avila-Ortiz et al., 2020 ; 2017; Sun et al., 2019 ). Seven studies assessed horizontal changes Barone et al., 2008 ; Canellas et al., 2020 ; Cha et al., 2019 ; Clementini at different levels (i.e. 1, 3 and 5 mm apical to the crestal bone) et al., 2019 ; Festa et al., 2013 ; Hauser et al., 2013 ; Jiang et al., 2017 ; (Canellas et al., 2020 ; Cha et al., 2019 ; Clementini et al., 2019; Jung Jung et al., 2013 ; Karaca et al., 2015 ; Lim et al., 2019 ; Pelegrine et al., 2013 ; Lim et al., 2019 ; Qabbani et al., 2017 ; Sun et al., 2019 ). et al., 2010; Qabbani et al., 2017 ; Sbordone et al., 2017 ; Sisti et al., More horizontal resorption in molar sites compared with non-molar 2012; Spinato et al., 2014 ; Sun et al., 2019 ; Thalmair et al., 2013 ; sites was consistently reported. Thoma et al., 2020 ; Zhao et al., 2018 ) and five studies did not specify whether complications occurred (Barone et al., 2017 ; Iasella et al., Radiographic vertical bone changes 2003; Machtei et al., 2019 ; Pang et al., 2016 ; Rasperini et al., 2010 ). Differences in vertical bone changes between baseline and final follow-up assessed in CBCT scan images were reported in ten studies (Araujo et al., 2015 ; Avila-Ortiz et al., 2020 ; Canellas et al., 3 . 4 . 3 | Feasibility of implant placement 2020; Clementini et al., 2019; Fiorellini et al., 2005 ; Jung et al., 2013 ; Karaca et al., 2015 ; Lim et al., 2019 ; Sun et al., 2019 ; Zhao Ten studies reported data on the feasibility of implant placement et al., 2018). All studies reported changes at mid-facial and mid-lin- with no additional bone grafting procedures at the same time or gual sites, except for one (Fiorellini et al., 2005) that only reported prior to implant placement (Avila-Ortiz et al., 2020 ; Cha et al., mid-facial vertical bone changes. Overall, more bone resorption at 2019; Fiorellini et al., 2005 ; Hauser et al., 2013 ; Jiang et al., 2017 ; mid-facial sites compared with mid-lingual sites was reported in the Lim et al., 2019 ; Rasperini et al., 2010 ; Sisti et al., 2012 ; Sun et al., selected studies. Non-molar sites exhibited less vertical bone loss 2019; Thoma et al., 2020 ). Only two of the selected trials reported than molar sites. the criteria applied to make this determination (Avila-Ortiz et al., 2020; Thoma et al., 2020 ). Six of them reported data exclusively Volumetric bone changes assessed in DICOM files on non-molar sites (Avila-Ortiz et al., 2020 ; Fiorellini et al., 2005 ; Differences in bone volumetric changes between baseline and final Hauser et al., 2013 ; Jiang et al., 2017 ; Sisti et al., 2012 ; Thoma follow-up assessed by CBCT scan images were reported in two stud- et al., 2020 ), three studies on molar sites (Cha et al., 2019 ; Lim ies including non-molar teeth (Avila-Ortiz et al., 2020 ; Canellas et al., et al., 2019; Rasperini et al., 2010 ) and one study on non-molar and 2020 ). The trial by Canellas et al. revealed less volumetric reduction molar sites (Sun et al., 2019 ). Overall, these studies showed that but a higher standard deviation than the study by Avila-Ortiz et al. 69.7% of non-molar and 45.9% of molar sites required additional (−44.87 ± 200.20 vs −154.51 ± 69.35 mm3 ). bone augmentation. COUSO-QUEIRUGA ET AL. | 139

FIGURE 4 Visual representation of the main findings derived from the quantitative analyses

3 . 4 . 4 | Patient-reported outcomes reduction was observed when only maxillary sites were analysed (p < 0.00001, MD: 2.38 mm, 95% CI: 1.46–3.29, I 2 = 97.0%). Less Only two studies reported PROMS (Avila-Ortiz et al., 2020 ; Machtei horizontal bone resorption in non-molar maxillary sites was ob- et al., 2019). The study by Machtei used a VAS scale with a range be- served when a thick facial bone plate was present (p < 0.00001, MD: tween 0 and 5 to assess the post-operative pain level at the 2 weeks 1.17 mm, 95% CI: 0.84–1.50) compared to sites with thin facial bone post-op visit. This trial reported higher mean values than the study (p < 0.00001, MD: 2.67 mm, 95% CI: 2.26–3.08). by Avila-Ortiz and co-workers in which discomfort was analysed at each post-operative visit using a VAS scale between 0 and 100, showing a progressive decrease over time. This was the only study 3 . 5 . 2 | Clinical mid-facial vertical bone changes reporting the perceived benefit of therapy after completion of treat- in non-molar sites ment (95.52 ± 9.56 out of 100). Evidence of mid-facial vertical bone reduction from baseline to final follow-up in non-molar sites was found (p < 0.00001, MD: 3 . 5 | Quantitative analyses (Pooled estimates) 1.71 mm, 95% CI: 1.30–2.12, I 2 = 94.0%) (Figure S2 ). The magnitude of this dimensional change was lower when only maxillary Meta-analyses were only performed if at least two studies with sites were analysed (p < 0.00001, MD: 1.09 mm, 95% CI: 0.89– similar extraction site features reported the same outcome of inter- 1.29, I 2 = 69.0%). Less mid-facial bone height loss was observed est. As shown in Figure 1 , twenty studies met this criterion (Aimetti in non-molar maxillary sites presenting a thick facial bone plate et al., 2009; Araujo et al., 2015 ; Avila-Ortiz et al., 2020 ; Barone et al., (p < 0.00001, MD: 0.50 mm, 95% CI: 0.07–0.93) compared to sites 2008 , 2017; Canellas et al., 2020; Cha et al., 2019 ; Clementini et al., with a thin facial bone plate (p < 0.00001, MD: 1.17 mm, 95% CI: 2019; Festa et al., 2013 ; Iasella et al., 2003 ; Jiang et al., 2017 ; Jung 0.86–1.48). et al., 2013 ; Lim et al., 2019 ; Machtei et al., 2019 ; Pang et al., 2016 ; Pelegrine et al., 2010 ; Qabbani et al., 2017 ; Spinato et al., 2014 ; Sun et al., 2019; Zhao et al., 2018 ). Available data allowed for grouping by 3 . 5 . 3 | Clinical mid-lingual vertical bone changes arch (i.e. maxillary, mandibular or both) and location (i.e. non-molar in non-molar sites or molar sites). A visual summary of the main findings is displayed in Figure 4 . Evidence of mid-lingual vertical bone reduction from baseline to final follow-up in non-molar sites was found (p < 0.00001, MD: 1.44 mm, 95% CI: 0.78–2.10, I 2 = 91.0%) (Figure S3). When only 3 . 5 . 1 | Clinical horizontal bone changes in non- maxillary sites were analysed, less bone reduction was observed molar sites (p < 0.00001, MD: 0.79 mm, 95% CI: 0.50–1.08, I 2 = 27.0%). Maxillary sites with thick facial bone showed less mid-lingual re- Evidence of horizontal bone reduction from baseline to final follow- sorption (p = 0.02, MD: 0.50 mm, 95% CI: 0.07–0.93) compared up in non-molar sites was found (p < 0.00001, MD: 2.73 mm, 95% with those exhibiting a thin facial bone plate (p = 0.0001, MD: CI: 2.36–3.11, I 2 = 94.0%) (Figure S1 ). Slightly less horizontal bone 1.00 mm, 95% CI: 0.49–1.51). 140 | COUSO-QUEIRUGA ET AL.

3 . 5 . 4 | Radiographic horizontal bone changes 2010; Qabbani et al., 2017 ; Sbordone et al., 2017 ; Sisti et al., 2012 ; in non-molar and molar sites Spinato et al., 2014 ; Sun et al., 2019 ; Thalmair et al., 2013 ; Thoma et al., 2020; Zhao et al., 2018 ). With respect to pooled estimates Evidence of horizontal bone reduction from baseline to final follow- (i.e., meta-analyses), these revealed that the magnitude of post-ex- up was found in non-molar sites (p < 0.00001, MD: 2.54 mm, 95% CI: traction horizontal changes in molar sites (assessed radiographically) 1.97–3.11, I 2 = 91.0%) and molar sites (p < 0.00001, MD: 3.61 mm, was substantially higher compared with non-molar sites (assessed 95% CI: 3.24–3.98, I 2 = 80.0%) (Figure S4 ). The magnitude of hori- clinically or radiographically), except for vertical mid-facial changes, zontal bone loss was higher in molar sites vs non-molar sites (3.61 vs as shown in Figure 4 . Data regarding facial bone thickness indicated 2.54 mm, respectively). that it can be used as a predictor of the extent of alveolar bone resorption in non-molar sites, as shown in Figures S1 , S2 and S3.

3 . 5 . 5 | Radiographic mid-facial vertical bone changes in non-molar and molar sites 4 . 2 | Quality of evidence

Evidence of mid-facial vertical bone reduction from baseline to final The overall quality of evidence of the included studies was average follow-up was found in non-molar sites (p < 0.00001, MD: 1.65 mm, to low. Only five of the included RCTs met all criteria of the Cochrane 95% CI: 0.42–2.88, I 2 = 96.0%) and molar sites (p < 0.0001, MD: Collaboration' s tool for assessing risk of bias (Avila-Ortiz et al., 2020 ; 1.46 mm, 95% CI: 0.73–2.20, I 2 = 30.0%) (Figure S5 ). The magnitude Barone et al., 2017 ; Canellas et al., 2020 ; Clementini et al., 2019; of mid-facial vertical bone resorption was slightly higher in non-mo- Sisti et al., 2012 ). Sixteen studies were categorized as presenting low lar sites (1.65 vs 1.46 mm). risk, fourteen of which were RCTs (Araujo et al., 2015 ; Avila-Ortiz et al., 2020; Barone et al., 2008 , 2017; Canellas et al., 2020 ; Cha et al., 2019; Clementini et al., 2019; Iasella et al., 2003 ; Jung et al., 3 . 5 . 6 | Radiographic mid-lingual vertical bone 2013 ; Lim et al., 2019 ; Machtei et al., 2019 ; Rasperini et al., 2010 ; changes in non-molar and molar sites Sisti et al., 2012 ; Sun et al., 2019 ) and two were non-RCTs (Sbordone et al., 2017; Zhao et al., 2018 ). Two RCTs were categorized as pre- Evidence of mid-lingual vertical bone reduction from baseline to senting some concerns (Hauser et al., 2013 ; Pang et al., 2016 ), nine final follow-up was found in non-molar sites (p < 0.00001, MD: RCTs were categorized as presenting high risk (Aimetti et al., 2009 ; 0.87 mm, 95% CI: 0.36–1.38, I 2 = 97.0%) and molar sites (p < 0.0001, Festa et al., 2013 ; Fiorellini et al., 2005 ; Karaca et al., 2015 ; Pelegrine MD: 1.20 mm, 95% CI: 0.56–1.83, I 2 = 53%) (Figure S6 ). Less mid- et al., 2010; Qabbani et al., 2017 ; Spinato et al., 2014 ; Thalmair et al., lingual vertical bone loss was observed in non-molar vs molar sites 2013 ; Thoma et al., 2020 ) and one non-RCT was categorized as pre- (0.79 vs 1.20 mm). senting serious risk (Jiang et al., 2017 ). Hence, the findings from this SR should be interpreted with caution.

4 | DISCUSSION 4 . 3 | Potential biases in the review process 4 . 1 | Summary of main findings In spite of having adhered to high methodological standards, this Data collected in this SR indicate that the naturally occurring resorp- review is not exempt of limitations. Aside from facial bone thick- tive process that follows tooth extraction results into significant ness, the effect of local, systemic and surgical factors could not changes to the alveolar ridge dimensions, irrespective of tooth type. be assessed as part of the quantitative analyses. This was mainly Molar sites generally exhibit more resorption on the coronal aspect due to discrepancies between studies or missing data. For exam- of the alveolar ridge for all dimensions analysed, except for vertical ple, local factors, such as whether socket integrity influences the mid-facial changes compared with non-molar sites. Additionally, a alveolar ridge dimensions, could not be assessed due to the wide combined estimation of the data from studies reporting the need methodological heterogeneity in the selected trials and the lack for bone augmentation to facilitate implant placement revealed that of detail in the available data. Baseline KTW was only reported that 69.7% of non-molar sites and 45.9% of molar sites required in four studies (Avila-Ortiz et al., 2020 ; Clementini et al., 2019; additional bone augmentation. Although complications were only Lim et al., 2019 ; Zhao et al., 2018 ). Soft tissue thickness was only described in one study (Fiorellini et al., 2005 ), most of the studies reported in two trials (Avila-Ortiz et al., 2020 ; Clementini et al., (n = 22) reported uneventful healing after tooth extraction (Aimetti 2019). Data regarding history of periodontitis were only reported et al., 2009; Araujo et al., 2015 ; Avila-Ortiz et al., 2020 ; Barone in one study (Zhao et al., 2018 ). Supracrestal tissue attachment et al., 2008 ; Canellas et al., 2020 ; Cha et al., 2019 ; Clementini et al., dimensions were not reported in any of the selected articles. 2019; Festa et al., 2013 ; Hauser et al., 2013 ; Jiang et al., 2017 ; Jung Systemic factors such as smoking and diabetes could not be as- et al., 2013 ; Karaca et al., 2015 ; Lim et al., 2019 ; Pelegrine et al., sessed due to the small sample size, methodological discrepancies, COUSO-QUEIRUGA ET AL. | 141 lack of detailed information and vague eligibility criteria in the in- (Chappuis et al., 2013 ). In this study, it was observed that thin facial cluded studies. Finally, although reported in several studies, the bone (<1 mm) was associated with more bone resorption than thick effect of surgical variables (flap elevation and primary vs second- bone (≥ 1 mm). Furthermore, a trial involving 59 patients revealed ary intention healing) on dimensional changes could not be quan- that the thicker the facial bone, the less the bone volume loss that titatively analysed since they were not consistently reported in occurs after a 14-week healing period (Avila-Ortiz et al., 2020 ). the trials that provided comparable data on the primary outcomes The lower need for bone augmentation prior to or simultaneously of interest (i.e. linear horizontal, mid-facial and mid-lingual ridge with implant placement in molar sites identified in the present SR reduction). Likewise, STL file analyses were only reported in five could be explained by the wider horizontal dimension of these sites studies. Linear soft tissue contour changes were evaluated in two which, in spite of undergoing more bone loss, allow for regular im- studies (Thalmair et al., 2013 ; Thoma et al., 2020 ), while volu- plant placement, compared with non-molar sites, where the impact metric soft tissue contour changes were assessed in three stud- of physiologic bone resorption is proportionally larger. This finding ies (Avila-Ortiz et al., 2020; Sbordone et al., 2017 ; Thalmair et al., is in agreement with the results published in a previous SR (Avila- 2013). Radiographic bone volume assessments were reported Ortiz et al., 2019 ), in which data from non-molar and molar sites in two studies (Avila-Ortiz et al., 2020 ; Canellas et al., 2020 ). were combined, that reported that 33.3%–66% of the untreated ex- PROMs were reported in two studies (Avila-Ortiz et al., 2020 ; traction sites required ancillary bone grafting prior to or at the time Machtei et al., 2019). In the light of the high degree of methodo- of implant placement. Interestingly, one trial including only maxillary logical heterogeneity across studies, the aforementioned second- molar sites revealed that 100% of the patients needed sinus aug- ary outcomes could not be further evaluated in a pooled analysis. mentation procedures for implant placement purposes (Cha et al., Finally, data from sites that received any further therapy beyond 2019). However, there was a lack of information regarding whether standard-of-care tooth extraction (e.g. collagen plug, autologous bone grafting augmentation procedures were needed to compen- blood-derived products, bone grafting, socket sealing, immediate sate the alveolar ridge resorption following tooth extraction, or implant placement or delivery of a removable prosthesis) were not whether there was a pre-existing vertical height deficiency at base- included in this review. Although it may be argued that the effect line. Therefore, this study was excluded from the analysis pertaining of, for example, a rapidly absorbable collagen sponge on dimen- to this parameter due to unclear description. This contrasts with the sional changes is negligible, this was done to homogenize the clini- findings from a study conducted by Rasperini et al. (2010 ) who re- cal scenario of interest as much as possible with the purpose of ported that only 37.5% maxillary molar sites required sinus augmen- eliminating the effect of any additional intervention on the healing tation procedures after tooth extraction, and this requirement was sequence and, therefore, the outcomes of interest considered in more frequent in partially damaged sockets. It is important to note this review. that this outcome is associated with a high degree of methodological variability since bone height and width requirements may vary significantly depending on implant dimension (i.e. length and diameter) 4 . 4 | Agreements and disagreements with other preferences. studies or reviews

To the best of the authors’ knowledge, this SR represents the most 5 | CONCLUSIONS comprehensive analysis of the evidence to date regarding the extent of post-extraction alveolar ridge dimensional changes after un- 1 . Based on the qualitative and quantitative analyses performed assisted socket healing. In general, the results of the quantitative as part of this SR, it can be concluded that more reduction analyses conducted in this study are not in alignment with those in the alveolar ridge dimensions can be expected in molar reported in previous SRs on this topic (Tan et al., 2012 ; Van der sites in all dimensions, except for vertical mid-facial changes, Weijden et al., 2009 ). This could be explained by methodological dis- compared with non-molar sites after a variable post-extraction crepancies, such as eligibility criteria, number and characteristics of healing period ranging from 2 to 9 months. selected articles, detail of data collection and also that previous SRs 2 . Ridge resorption is more pronounced in the horizontal dimension, pooled non-molar and molar sites and did not differentiate between followed by vertical mid-facial and mid-lingual changes. clinical and radiographic assessments. 3 . Facial bone thickness upon extraction seems to be strongly asso- The findings of this SR regarding horizontal ridge reduction in ciated with the extent and magnitude of alveolar bone resorption. molar sites are in accordance with those from a classic clinical study The thicker the facial bone, the less the ridge resorption. in which a higher per cent of horizontal ridge resorption was ob- 4 . According to the current base of evidence, non-molar sites are served in maxillary and mandibular molar regions compared with associated with an increased need for ancillary bone grafting pro- premolar sites (Schropp et al., 2003 ). cedures prior to or at the time of implant placement compared Findings pertaining to the predictive value of facial bone thick- with molar sites (69.7% vs 45.9%, respectively). ness are in accordance with the observations reported in a case-se- 5 . Future studies should improve the report of information on local ries study that involved the extraction of 39 maxillary anterior teeth (i.e. socket anatomy and integrity, soft tissue thickness, keratinized 142 | COUSO-QUEIRUGA ET AL.

mucosa width, supracrestal tissue height), systemic (i.e. diabetes, Research, 93 ( 10 ), 950 – 958. https://doi.org/10.1177/00220 34514 541127 smoking status, history of periodontitis) and surgical variables (i.e. Avila-Ortiz, G. , Gubler, M. , Romero-Bustillos, M. , Nicholas , C. L. , flap elevation, primary closure) on the dimensional changes in the Zimmerman, M. B. , & Barwacz, C. A. (2020 ). Efficacy of alveolar alveolar ridge that follow unassisted tooth extraction. ridge preservation: A randomized controlled trial. Journal of Dental Research, 99 ( 4 ), 402– 409 . https://doi.org/10.1177/00220 34520 905660 C O N F L I C T O F I N T E R E S T Barone, A. , Aldini, N. N., Fini, M. , Giardino, R. , Calvo Guirado, J. L. , & The authors have no conflicts of interest to report pertaining to the Covani, U. ( 2008 ). Xenograft versus extraction alone for ridge conduction of this SRs. This study was supported by the University preservation after tooth removal: A clinical and histomorphomet- of Iowa College of Dentistry Department of Periodontics Graduate ric study. Journal of Periodontology, 79( 8 ), 1370– 1377 . https://doi. Student Research Fund. org/10.1902/jop.2008.070628 Barone, A. , Toti, P. , Quaranta, A. , Alfonsi, F. , Cucchi, A. , Negri, B. , Di Felice, R. , Marchionni, S. , Calvo-Guirado, J. L. , Covani, U. , & Nannmark, U. AUTHOR CONTRIBUTIONS (2017 ). Clinical and histological changes after ridge preservation E.C.Q. and G.A.O. conceived the idea; M.T. and E.C.Q. screened the with two xenografts: Preliminary results from a multicentre ran- initial entries, selected the articles and collected the data. S.S. and domized controlled clinical trial. Journal of Clinical Periodontology, 44 (2 ), 204 –214 . https://doi.org/10.1111/jcpe.12655 M.T. assessed the risk of bias. L.C contributed to the design and ana- Barootchi, S. , Wang, H. L. , Ravida, A. , Ben Amor, F. , Riccitiello, F. , Rengo, lysed the data. E.C.Q. and G.A.O. led the writing. S.S. and L.C. criti- C. , & Sammartino, G. (2019 ). Ridge preservation techniques to avoid cally revised the manuscript. invasive bone reconstruction: A systematic review and meta-analysis: Naples Consensus Report Working Group C. International Journal of Oral Implantology, 12 ( 4 ), 399 – 416 . DATA AVAILABILITY Bassir, S. H. , Alhareky, M. , Wangsrimongkol, B. , Jia, Y. , & Karimbux, Data sets are available from the corresponding author upon reason- N. (2018 ). Systematic review and meta-analysis of hard tissue able request. outcomes of alveolar ridge preservation. International Journal of Oral and Maxillofacial Implants, 33 ( 5), 979 – 994 . https://doi. org/10.11607/ jomi.6399 O R C I D Canellas , J. V. D. S. , da Costa, R. C. , Breves, R. C. , de Oliveira, G. P. , E m i l i o C o u s o - Q u e i r u g a https://orcid.org/0000-0002-9989-4483 Figueredo, C. M. D. S. , Fischer, R. G. , Thole, A. A. , Medeiros, P. J. D' . Sandra Stuhr https://orcid.org/0000-0003-3473-2149 A. , & Ritto, F. G. ( 2020). Tomographic and histomorphometric eval- Mustafa Tattan https://orcid.org/0000-0001-7498-8064 uation of socket healing after tooth extraction using leukocyte- and platelet-rich fibrin: A randomized, single-blind, controlled clinical Leandro Chambrone https://orcid.org/0000-0002-2838-1015 trial . Journal of Cranio-Maxillo-Facial Surgery, 48( 1), 24 –32 . https:// Gustavo Avila-Ortiz https://orcid.org/0000-0002-5763-0201 doi.org/10.1016/j.jcms.2019.11.006 Canellas , J. , Ritto, F. G. , Figueredo, C. , Fischer, R. G. , de Oliveira, G. P. , REFERENCES Thole, A. A. , & Medeiros, P. J. D. (2019 ). Histomorphometric evaluation of different grafting materials used for alveolar ridge preserva- Aimetti, M. , Romano, F. , Griga, F. B. , & Godio, L. ( 2009). Clinical and his- tion: A systematic review and network meta-analysis. International tologic healing of human extraction sockets filled with calcium sul- Journal of Oral and Maxillofacial Surgery, https://doi.org/10.1016/j. fate. International Journal of Oral and Maxillofacial Implants, 24 ( 5), ijom.2019.10.007 902 – 909 . Cardaropoli, G. , Araujo, M. , & Lindhe, J. ( 2003 ). Dynamics of bone tis- Al Yafi, F. , Alchawaf, B. , & Nelson, K. (2019 ). What is the optimum for sue formation in tooth extraction sites. An experimental study in alveolar ridge preservation? Dental Clinics of North America, 63( 3 ), dogs. Journal of Clinical Periodontology, 30 ( 9 ), 809 – 818 . https://doi. 399– 418 . https://doi.org/10.1016/j.cden.2019.02.007 org/10.1034/j.1600-051x.2003.00366.x Annunziata, M. , Guida, L. , Nastri, L. , Piccirillo, A. , Sommese, L. , & Cha, J. K. , Song, Y. W. , Park, S. H. , Jung, R. E. , Jung, U. W. , & Thoma, D. Napoli, C. (2018 ). The role of autologous platelet concen- S. (2019 ). Alveolar ridge preservation in the posterior maxilla re- trates in alveolar socket preservation: A systematic review. duces vertical dimensional change: A randomized controlled clini- Transfusion Medicine and Hemotherapy, 45 (3 ), 195– 203. https:// cal trial. Clinical Oral Implants Research, 30 (6 ), 515 – 523 . https://doi. doi.org/10.1159/00048 8061 org/10.1111/clr.13436 Araujo, M. G. , da Silva , J. C. C. , de Mendonca , A. F. , & Lindhe, J. ( 2015). Chan, H. L. , Lin, G. H. , Fu, J. H. , & Wang, H. L. (2013 ). Alterations in bone Ridge alterations following grafting of fresh extraction sockets in quality after socket preservation with grafting materials: A system- man. A randomized clinical trial. Clinical Oral Implants Research, atic review. International Journal of Oral and Maxillofacial Implants, 26( 4 ), 407 –412 . https://doi.org/10.1111/clr.12366 28 (3 ), 710 – 720 . https://doi.org/10.11607/jomi.2 913 Araujo, M. G. , & Lindhe, J. ( 2005). Dimensional ridge alterations fol- Chapple, I. L. , & Wilson, N. H. (2014 ). Manifesto for a paradigm shift: lowing tooth extraction. An experimental study in the dog. periodontal health for a better life. British Dental Journal, 216 ( 4 ), Journal of Clinical Periodontology, 32 (2 ), 212 – 218 . https://doi. 159 – 162 . https://doi.org/10.1038/sj.bdj.2014.97 org/10.1111/j.1600-051X.2005.00642.x Chappuis, V. , Engel, O. , Reyes , M. , Shahim, K. , Nolte , L. P. , & Buser, D. Avila-Ortiz, G. , Chambrone, L. , & Vignoletti, F. (2019 ). Effect of alve- (2013 ). Ridge alterations post-extraction in the esthetic zone: A 3D olar ridge preservation interventions following tooth extraction: analysis with CBCT. Journal of Dental Research, 92 ( 12 Suppl), 195S – A systematic review and meta-analysis. Journal of Clinical 201S. https://doi.org/10.1177/0022034513 506713 Periodontology, 46( Suppl 21), 195– 223. https://doi.org/10.1111/ Chappuis, V. , Engel, O. , Shahim, K. , Reyes , M. , Katsaros, C. , & Buser, jcpe.13057 D. (2015 ). Soft tissue alterations in esthetic postextraction sites: Avila-Ortiz, G. , Elangovan, S. , Kramer, K. W. , Blanchette, D. , & Dawson, A 3-dimensional analysis. Journal of Dental Research, 94 ( 9 Suppl), D. V. ( 2014). Effect of alveolar ridge preservation after tooth ex- 187S – 193S. https://doi.org/10.1177/0022034515 592869 traction: a systematic review and meta-analysis. Journal of Dental COUSO-QUEIRUGA ET AL. | 143

Clementini, M. , Agostinelli, A. , Castelluzzo, W. , Cugnata, F. , Vignoletti, Jambhekar, S. , Kernen, F. , & Bidra, A. S. ( 2015). Clinical and histologic F. , & De Sanctis, M. (2019 ). The effect of immediate implant place- outcomes of socket grafting after flapless tooth extraction: A sys- ment on alveolar ridge preservation compared to spontaneous tematic review of randomized controlled clinical trials. Journal of healing after tooth extraction: Radiographic results of a random- Prosthetic Dentistry, 113 ( 5), 371– 382 . https://doi.org/10.1016/j. ized controlled clinical trial . Journal of Clinical Periodontology, 46 (7 ), prosdent.2014.12.009 776– 786 . https://doi.org/10.1111/jcpe.13125 Jiang, X. , Zhang , Y. , Chen , B. , & Lin, Y. (2017 ). Pressure bearing device De Risi, V. , Clementini, M. , Vittorini, G. , Mannocci, A. , & De Sanctis, affects extraction socket remodeling of maxillary anterior tooth. M. (2015 ). Alveolar ridge preservation techniques: A systematic A prospective clinical trial. Clinical Implant Dentistry and Related review and meta-analysis of histological and histomorphometri- Research, 19 (2 ), 296 – 305 . https://doi.org/10.1111/cid.12442 cal data. Clinical Oral Implants Research, 26( 1), 50– 68 . https://doi. Jung, R. E. , Philipp, A. , Annen, B. M. , Signorelli , L. , Thoma, D. S. , org/10.1111/clr.12288 Hämmerle, C. H. F. , Attin, T. , & Schmidlin, P. (2013 ). Radiographic Del Fabbro, M. , Bucchi, C. , Lolato, A. , Corbella , S. , Testori, T. , & Taschieri, evaluation of different techniques for ridge preservation after tooth S. ( 2017). Healing of postextraction sockets preserved with autolo- extraction: a randomized controlled clinical trial. Journal of Clinical gous platelet concentrates. A systematic review and meta-analysis. Periodontology, 40( 1), 90 – 98. https://doi.org/10.1111/jcpe.12027 Journal of Oral and Maxillofacial Surgery, 75( 8 ), 1601– 1615. https:// Kao, R. T. (2008 ). Strategic extraction: a paradigm shift that is changing doi.org/10.1016/j.joms.2017.02.009 our profession. Journal of Periodontology, 79 ( 6 ), 971 – 977. https:// Discepoli, N. , Vignoletti, F. , Laino, L. , de Sanctis , M. , Munoz, F. , & doi.org/10.1902/jop.2008.070551 Sanz, M. (2013 ). Early healing of the alveolar process after tooth Karaca, C. , Er, N. , Gulsahi, A. , & Koseoglu, O. T. ( 2015). Alveolar ridge extraction: an experimental study in the beagle dog. Journal of preservation with a free gingival graft in the anterior maxilla: Clinical Periodontology, 40 ( 6 ), 638 – 644 . https://doi.org/10.1111/ Volumetric evaluation in a randomized clinical trial. International jcpe.12074 Journal of Oral and Maxillofacial Surgery, 44 (6 ), 774– 780 . https://doi. Evian, C. I. , Rosenberg, E. S. , Coslet, J. G. , & Corn, H. ( 1982 ). The os- org/10.1016/j.ijom.2015.01.015 teogenic activity of bone removed from healing extraction sock- Lee , J. , Lee , J. B. , Koo, K. T. , Seol, Y. J. , & Lee , Y. M. (2018 ). Flap manage- ets in humans. Journal of Periodontology, 53 ( 2 ), 81 – 85. https://doi. ment in alveolar ridge preservation: A systematic review and me- org/10.1902/jop.1982.53.2.81 ta-analysis. International Journal of Oral and Maxillofacial Implants, Faria-Almeida, R. , Astramskaite-Januseviciene, I. , Puisys, A. , & Correia, F. 33( 3 ), 613 – 621 . https://doi.org/10.11607/jomi.6368 (2019 ). Extraction socket preservation with or without membranes, Lim, H. C. , Shin, H. S. , Cho, I. W. , Koo, K. T. , & Park, J. C. ( 2019). Ridge soft tissue influence on post extraction alveolar ridge preservation: preservation in molar extraction sites with an open-healing ap- A systematic review. Journal of Oral and Maxillofacial, 10 (3 ), e5 . proach: A randomized controlled clinical trial. Journal of Clinical https://doi.org/10.5037/jomr.2019.10305 Periodontology, 46 ( 11 ), 1144– 1154 . https://doi.org/10.1111/ Festa, V. M. , Addabbo, F. , Laino, L. , Femiano, F. , & Rullo, R. (2013 ). Porcine- jcpe.13184 derived xenograft combined with a soft cortical membrane versus MacBeth, N. , Trullenque-Eriksson, A. , Donos, N. , & Mardas , N. (2017 ). extraction alone for implant site development: A clinical study in Hard and soft tissue changes following alveolar ridge preservation: humans . Clinical Implant Dentistry and Related Research, 15 ( 5), 707 – A systematic review. Clinical Oral Implants Research, 28 (8 ), 982– 713. https://doi.org/10.1111/j.1708-8208.2011.00398.x 1004. https://doi.org/10.1111/clr.12911 Fiorellini, J. P. , Howell, T. H. , Cochran, D. , Malmquist, J. , Lilly, L. C., Machtei, E. E. , Mayer, Y. , Horwitz, J. , & Zigdon-Giladi, H. (2019 ). Spagnoli, D. , Toljanic, J. , Jones, A. , & Nevins, M. (2005 ). Randomized Prospective randomized controlled clinical trial to compare hard study evaluating recombinant human bone morphogenetic pro- tissue changes following socket preservation using alloplasts, xe- tein-2 for extraction socket augmentation. Journal of Periodontology, nografts vs no grafting: Clinical and histological findings. Clinical 76( 4 ), 605 – 613 . https://doi.org/10.1902/jop.2005.76.4.605 Implant Dentistry and Related Research, 21( 1), 14– 20 . https://doi. Hauser, F. , Gaydarov, N. , Badoud, I. , Vazquez, L. , Bernard, J. P. , & org/10.1111/cid.12707 Ammann, P. ( 2013). Clinical and histological evaluation of postex- Majzoub, J. , Ravida, A. , Starch-Jensen, T. , Tattan, M. , & Suarez-Lopez traction platelet-rich fibrin socket filling: A prospective randomized Del Amo, F. ( 2019). The influence of different grafting materials controlled study. Implant Dentistry, 22 ( 3 ), 295 – 303 . https://doi. on alveolar ridge preservation: A systematic review. Journal of org/10.1097/ID.0b013 e3182906eb3 Oral and Maxillofacial Research, 10 (3 ), e6 . https://doi.org/10.5037/ Higgins, J. P. T. , Savović , J. , Page, M. J. , Elbers, R. G. , & Sterne, J. A. C. jomr.2019.10306 (2019 ). Assessing risk of bias in a randomized trial. In J. P. T. Higgins Mardas , N. , Trullenque-Eriksson, A. , MacBeth, N. , Petrie, A. , & Donos, & J. Thomas (Eds.), Cochrane handbook for systematic reviews of in- N. ( 2015). Does ridge preservation following tooth extraction im- terventions (pp. 205 – 228 ). Cochrane. https://www.train ing.cochr prove implant treatment outcomes: A systematic review: Group 4: ane.org/handbook Therapeutic concepts & methods. Clinical Oral Implants Research, Horvath, A. , Mardas , N. , Mezzomo, L. A. , Needleman, I. G. , & Donos, 26( Suppl 11), 180 –201 . https://doi.org/10.1111/clr.12639 N. (2013 ). Alveolar ridge preservation. A systematic review. Clinical Moher, D. , Liberati, A. , Tetzlaff, J. , Altman, D. G. , & PRISMA Group ( 2009). Oral Investigations, 17 ( 2 ), 341– 363 . https://doi.org/10.1007/s0078 Preferred reporting items for systematic reviews and meta-analy- 4-012-0758-5 ses: The PRISMA statement. Journal of Clinical Epidemiology, 62 ( 10 ), Iasella, J. M. , Greenwell, H. , Miller, R. L. , Hill, M. , Drisko, C. , Bohra , A. 1006 – 1012 . https://doi.org/10.1016/j.jclinepi.2009.06.005 A. , & Scheetz, J. P. ( 2003). Ridge preservation with freeze-dried Moraschini, V. , & Barboza, E. S. ( 2015). Effect of autologous platelet con- bone allograft and a collagen membrane compared to extraction centrates for alveolar socket preservation: A systematic review. alone for implant site development: a clinical and histologic study International Journal of Oral and Maxillofacial Surgery, 44 ( 5), 632 – in humans . Journal of Periodontology, 74 ( 7 ), 990 – 999 . https://doi. 641 . https://doi.org/10.1016/j.ijom.2014.12.010 org/10.1902/jop.2003.74.7.990 Moraschini, V. , & Barboza Edos, S. (2016 ). Quality assessment of system- Iocca, O. , Farcomeni, A. , Pardinas Lopez, S. , & Talib, H. S. (2017 ). Alveolar atic reviews on alveolar socket preservation . International Journal ridge preservation after tooth extraction: A Bayesian Network of Oral and Maxillofacial Surgery, 45 ( 9 ), 1126 – 1134. https://doi. meta-analysis of grafting materials efficacy on prevention of bone org/10.1016/j.ijom.2016.03.010 height and width reduction. Journal of Clinical Periodontology, 44 ( 1), Morjaria, K. R. , Wilson, R. , & Palmer, R. M. ( 2014). Bone heal- 104– 114 . https://doi.org/10.1111/jcpe.12633 ing after tooth extraction with or without an intervention: 144 | COUSO-QUEIRUGA ET AL.

A systematic review of randomized controlled trials. Clinical Boutron, I. , Carpenter, J. R. , Chan, A.-W. , Churchill, R. , Deeks, J. Implant Dentistry and Related Research, 16( 1), 1 –20 . https://doi. J. , Hróbjartsson, A. , Kirkham, J. , Jüni, P. , Loke, Y. K. , Pigott, T. D. , … org/10.1111/j.1708-8208.2012.00450.x Higgins, J. P. T. (2016 ). ROBINS-I: A tool for assessing risk of bias in Moslemi, N. , Khoshkam, V. , Rafiei, S. C. , Bahrami, N. , & Aslroosta, H. non-randomised studies of interventions. BMJ , 355 , i4919 . https:// (2018 ). Outcomes of alveolar ridge preservation with recom- doi.org/10.1136/bmj.i4919 binant human bone morphogenetic protein-2: A systematic re- Stumbras, A. , Kuliesius, P. , Januzis, G. , & Juodzbalys, G. ( 2019). Alveolar view. Implant Dentistry, 27 (3 ), 351 –362 . https://doi.org/10.1097/ ridge preservation after tooth extraction using different bone graft ID.0000000000 000722 materials and autologous platelet concentrates: A systematic re- Natto, Z. S. , Yaghmoor, W. , Bannuru, R. R. , & Nevins, M. (2017 ). view. Journal of Oral and Maxillofacial Research, 10 ( 1), e2. https:// Identification and efficacy ranking of allograft and xenograft for doi.org/10.5037/jomr.2019.10102 extraction and ridge preservation procedures. The International Sun, D. J. , Lim, H. C. , & Lee , D. W. (2019 ). Alveolar ridge preservation Journal of Periodontics and Restorative Dentistry, 37 ( 5), e253 – e260 . using an open membrane approach for sockets with bone de- https://doi.org/10.11607/prd.3323 ficiency: A randomized controlled clinical trial. Clinical Implant Pang , C. , Ding , Y. , Hu, K. , Zhou, H. , Qin, R. , & Hou, R. (2016 ). Influence Dentistry and Related Research, 21 ( 1), 175 – 182. https://doi. of preservation of the alveolar ridge on delayed implants after ex- org/10.1111/cid.12668 traction of teeth with different defects in the buccal bone. British Tan, W. L. , Wong, T. L. , Wong, M. C. , & Lang, N. P. ( 2012 ). A systematic Journal of Oral and Maxillofacial Surgery, 54( 2 ), 176 – 180. https://doi. review of post-extractional alveolar hard and soft tissue dimen- org/10.1016/j.bjoms.2015.11.025 sional changes in humans. Clinical Oral Implants Research, 23 ( Suppl Pelegrine, A. A. , da Costa , C. E. , Correa, M. E. , & Marques, J. F. 5), 1 –21 . https://doi.org/10.1111/j.1600-0501.2011.02375.x Jr (2010 ). Clinical and histomorphometric evaluation of ex- Ten Heggeler, J. M. , Slot, D. E. , & Van der Weijden, G. A. (2011 ). traction sockets treated with an autologous bone marrow graft. Effect of socket preservation therapies following tooth ex- Clinical Oral Implants Research, 21( 5), 535 – 542 . https://doi. traction in non-molar regions in humans: A systematic review. org/10.1111/j.1600-0501.2009.01891.x Clinical Oral Implants Research, 22 ( 8 ), 779 – 788 . https://doi. Pranskunas , M. , Galindo-Moreno, P. , & Padial-Molina, M. ( 2019). org/10.1111/j.1600-0501.2010.02064.x Extraction socket preservation using growth factors and stem Thalmair, T. , Fickl, S. , Schneider, D. , Hinze, M. , & Wachtel, H. ( 2013). cells: A systematic review. Journal of Oral and Maxillofacial Research, Dimensional alterations of extraction sites after different alveo- 10 (3 ), e7. https://doi.org/10.5037/jomr.2019.10307 lar ridge preservation techniques - a volumetric study. Journal of Qabbani, A. A. , Razak, N. H. A. , Kawas, S. A. , Sheikh Abdul Hamid, S. , Clinical Periodontology, 40 ( 7 ), 721 – 727. https://doi.org/10.1111/ Wahbi, S. , & Samsudin, A. R. (2017 ). The efficacy of immediate jcpe.12111 implant placement in extraction sockets for alveolar bone pres- Thoma, D. S. , Bienz, S. P. , Lim, H. C. , Lee , W. Z. , Hammerle, C. H. F. , & Jung, ervation: A clinical evaluation using three-dimensional cone beam R. E. (2020 ). Explorative randomized controlled study comparing computerized tomography and resonance frequency analysis soft tissue thickness, contour changes, and soft tissue handling of value. Journal of Craniofacial Surgery, 28 ( 4 ), e318 – e325. https://doi. two ridge preservation techniques and spontaneous healing two org/10.1097/SCS.0000000000 003569 months after tooth extraction. Clinical Oral Implants Research, 31( 6 ), Rasperini, G. , Canullo, L. , Dellavia, C. , Pellegrini, G. , & Simion, M. (2010 ). 565– 574 . https://doi.org/10.1111/clr.13594 Socket grafting in the posterior maxilla reduces the need for sinus Tonetti, M. S. , Steffen, P. , Muller-Campanile, V. , Suvan, J. , & Lang, N. P. augmentation. International Journal of Periodontics and Restorative (2000 ). Initial extractions and tooth loss during supportive care in Dentistry, 30 (3 ), 265 –273 . a periodontal population seeking comprehensive care. Journal of Sbordone, C. , Toti, P. , Martuscelli, R. , Guidetti, F. , Porzio , M. , & Sbordone, Clinical Periodontology, 27 ( 11 ), 824 – 831. https://doi.org/10.1034/ L. ( 2017). Evaluation of volumetric dimensional changes in posterior j.1600-051x.2000.027011824.x extraction sites with and without ARP using a novel imaging device. Troiano, G. , Zhurakivska, K. , Lo Muzio, L. , Laino, L. , Cicciu, M. , & Lo Russo, Clinical Implant Dentistry and Related Research, 19 (6 ), 1044 – 1053 . L. (2018 ). Combination of bone graft and resorbable membrane for https://doi.org/10.1111/cid.12537 alveolar ridge preservation: A systematic review, meta-analysis, Schropp, L. , Wenzel, A. , Kostopoulos, L. , & Karring, T. (2003 ). Bone and trial sequential analysis. Journal of Periodontology, 89 ( 1), 46 – 57 . healing and soft tissue contour changes following single-tooth ex- https://doi.org/10.1902/jop.2017.170241 traction: A clinical and radiographic 12-month prospective study. Trombelli, L. , Farina, R. , Marzola, A. , Bozzi, L. , Liljenberg, B. , & Lindhe, International Journal of Periodontics and Restorative Dentistry, 23 ( 4 ), J. (2008 ). Modeling and remodeling of human extraction sock- 313– 323 . ets. Journal of Clinical Periodontology, 35 (7 ), 630– 639 . https://doi. Sisti, A. , Canullo, L. , Mottola, M. P. , Covani, U. , Barone, A. , & org/10.1111/j.1600-051X.2008.01246.x Botticelli, D. ( 2012 ). Clinical evaluation of a ridge augmen- Van der Weijden, F. , Dell' Acqua , F. , & Slot, D. E. (2009 ). Alveolar bone tation procedure for the severely resorbed alveolar socket: dimensional changes of post-extraction sockets in humans: A sys- Multicenter randomized controlled trial, preliminary results. tematic review. Journal of Clinical Periodontology, 36 ( 12 ), 1048– Clinical Oral Implants Research, 23 ( 5), 526 –535 . https://doi. 1058 . https://doi.org/10.1111/j.1600-051X.2009.01482.x org/10.1111/j.1600-0501.2011.02386.x Vignoletti, F. , Matesanz, P. , Rodrigo, D. , Figuero, E. , Martin , C. , & Sanz, Spinato, S. , Galindo-Moreno, P. , Zaffe, D. , Bernardello, F. , & Soardi , C. M. ( 2012 ). Surgical protocols for ridge preservation after tooth ex- M. (2014 ). Is socket healing conditioned by buccal plate thickness? traction. A systematic review. Clinical Oral Implants Research, 23 ( Suppl A clinical and histologic study 4 months after mineralized human 5), 22 – 38 . https://doi.org/10.1111/j.1600-0501.2011.02331.x bone allografting . Clinical Oral Implants Research, 25( 2 ), e120– e126. Vittorini Orgeas, G. , Clementini, M. , De Risi, V. , & de Sanctis , M. ( 2013). https://doi.org/10.1111/clr.12073 Surgical techniques for alveolar socket preservation: A systematic Stedman, M. R. , Curtin , F. , Elbourne, D. R. , Kesselheim, A. S. , & review . International Journal of Oral and Maxillofacial Implants, 28 ( 4 ), Brookhart, M. A. (2011 ). Meta-analyses involving cross-over trials: 1049– 1061 . https://doi.org/10.11607/jomi.2670 Methodological issues. International Journal of Epidemiology, 40( 6 ), Willenbacher, M. , Al-Nawas, B. , Berres, M. , Kammerer, P. W. , & 1732 – 1734. https://doi.org/10.1093/ije/dyp345 Schiegnitz, E. (2016 ). The effects of alveolar ridge preservation: A Sterne, J. A. C. , Hernán, M. A. , Reeves , B. C. , Savović , J. , Berkman, meta-analysis. Clinical Implant Dentistry and Related Research, 18 (6 ), N. D. , Viswanathan, M. , Henry, D. , Altman, D. G. , Ansari , M. T. , 1248 – 1268. https://doi.org/10.1111/cid.12364 COUSO-QUEIRUGA ET AL. | 145

Zhao, L. , Xu, T. , Hu, W. , & Chung, K. H. ( 2018 ). Preservation and augmentation of molar extraction sites affected by severe bone defect How to cite this article: Couso-Queiruga E, Stuhr S, Tattan M, due to advanced periodontitis: A prospective clinical trial. Clinical Chambrone L, Avila-Ortiz G. Post-extraction dimensional Implant Dentistry and Related Research, 20 (3 ), 333– 344 . https://doi. org/10.1111/cid.12585 changes: A systematic review and meta-analysis. J Clin Periodontol. 2021;48:127–145. https://doi.org/10.1111/ jcpe.13390

S U P P O R T I N G I N F O R M A T I O N Additional supporting information may be found online in the Supporting Information section.

A P P E N D I X 1 To demonstrate the overall risk of bias, each included study was categorized as being at low, some concerns or high risk of bias ac- Search methods cording to the following criteria: The search strategy followed was: #1 Tooth extraction OR extraction OR dental extraction OR ex- 1 . Low risk of bias: the trial was judged to be at low risk of traction, tooth OR tooth socket bias for all domains for this result. #2 Alveolar ridge preservation OR ridge preservation OR socket 2 . Some concerns of bias: the trial was judged to raise some con- grafting OR socket filling OR socket preservation OR socket graft cerns in at least one domain for this result, but not to be at high OR guided bone regeneration OR alveolar ridge augmentation risk of bias for any domain. #3 #1 OR #2 3 . High risk of bias: the trial was judged to be at high risk of bias in at A complementary manual search of articles published in rel- least one domain for this result or have some concerns for multi- evant scientific journals (i.e. Journal of Clinical Periodontology, ple domains in a way that substantially lowers confidence in the Journal of Periodontology, International Journal of Periodontics result. and Restorative Dentistry, Clinical Implant Dentistry and Related Research, Clinical Oral Implants Research, Implant Dentistry, Journal The following domains were assessed for non-RCTs: of Oral Implantology, International Journal of Oral and Maxillofacial Implants, Journal of Oral and Maxillofacial Surgery, International 1 . Pre-intervention Journal of Oral Implantology, European Journal of Oral Implantology) • Bias due to confounding from 1 January 2020 to 15 August 2020 was performed. Additionally, • Bias in selection of participants into the study cross-referencing of cited references in 32 SRs on the topic published 2 . At intervention until 15 August 2020 was conducted (Al Yafi et al., 2019 ; Annunziata • Bias in classification of interventions et al., 2018 ; Avila-Ortiz et al., 2014 , 2019 ; Barootchi et al., 2019 ; 3 . Post-intervention Bassir et al., 2018 ; Canellas et al., 2019 ; Chan et al., 2013 ; De Risi • Bias due to deviations from intended interventions et al., 2015; Del Fabbro et al., 2017 ; Faria-Almeida et al., 2019 ; • Bias due to missing data Horvath et al., 2013 ; Iocca et al., 2017 ; Jambhekar et al., 2015 ; Lee • Bias in measurement of outcomes et al., 2018 ; MacBeth et al., 2017 ; Majzoub et al., 2019 ; Mardas et al., • Bias in selection of the reported result 2015; Moraschini & Barboza, 2015 ; Moraschini & Barboza Edos, To demonstrate the overall risk of bias, each included study was 2016 ; Morjaria et al., 2014 ; Moslemi et al., 2018 ; Natto et al., 2017 ; categorized as being low, moderate, serious and critical risk of bias or Pranskunas et al., 2019; Stumbras et al., 2019 ; Tan et al., 2012 ; Ten no information according to the following criteria: Heggeler et al., 20112011 ; Troiano et al., 2018 ; Van der Weijden et al., 2009; Vignoletti et al., 2012 ; Vittorini Orgeas et al., 2013 ; 1 . Low risk of bias: the study is judged to be at low risk of bias Willenbacher et al., 2016). for all domains. 2 . Moderate risk of bias: the study is judged to be at low or moderate Risk of bias risk of bias for all domains. The following domains were assessed for RCTs: 3 . Serious risk of bias: the study is judged to be at serious risk of bias in at least one domain, but not at critical risk of bias in any domain. 1 . Randomization process 4 . Critical risk of bias: the study is judged to be at critical risk of bias 2 . Deviations from intended interventions (assignment) in at least one domain. 3 . Missing outcome data 5 . No information: there is no clear indication that the study is at 4 . Measurement of the outcome serious or critical risk of bias, and there is a lack of information in 5 . Selection of the reported result one or more key domains of bias (a judgement is required for this). Received: 29 April 2020 | Revised: 3 September 2020 | Accepted: 6 October 2020 DOI: 10.1111/jcpe.13387

ORIGINAL ARTICLE IMPLANT DENTISTRY

Volumetric soft tissue alterations in the early healing phase after peri- implant soft tissue contour augmentation with a porcine collagen matrix versus the autologous connective tissue graft: A controlled clinical trial

1Department of Oral and Maxillofacial Surgery, University of Erlangen- Abstract Nuremberg, Erlangen, Germany Aim: This study evaluates the early volumetric changes after buccal soft tissue con- 2Private Practice, Stein, Germany tour augmentation around implants with a porcine collagen matrix (CM) vs. the sub- 3Private Practice, Munich, Germany epithelial connective tissue graft (SCTG) from the palate. 4Department of Medical Informatics, Biometry and Epidemiology (IMBE), Materials and methods: 14 patients were enrolled after early implant placement with University of Erlangen-Nuremberg, simultaneous contour augmentation and persistent buccal tissue deficits. At implant Erlangen, Germany 5Department of Prosthodontics, exposure, buccal soft tissues were thickened with the CM (n = 7) or the SCTG (n = 7). University of Erlangen-Nuremberg, Impressions were taken before and after surgery, after ten days, one, three and six Erlangen, Germany months. Impressions were digitized and augmented regions 3D evaluated (soft tissue Correspondence volume (mm3, %)/thickness (mm)). Christian M. Schmitt, Department of 3 3 Oral-and Maxillofacial Surgery, University Results: Volume increase (mm ) after 6 months was 19.56 ± 8.95 mm (CM) and of Erlangen- Nuremberg, Glückstrasse 11, 61.75 ± 52.69 mm3 (SCTG) (insignificant, p = .058). In percentage, this was a vol- 91054 Erlangen, Germany. Email: [email protected] ume loss of the initially augmented soft tissue volume (100%) of 81.76% in the CM group and 56.39% in the SCTG group (6 months). The mean soft tissue thickness Funding informationThis study was supported by a grant from Botiss Biomaterials increase (mm) in the buccal contour after 6 months was 0.30 ± 0.16 mm (CM) and GmbH, Zossen Germany. 0.80 ± 0.61 mm (SCTG) (insignificant, p = .071). Conclusion: The early healing phase is associated with a significant volume loss of the soft tissues. The SCTG shows insignificant superiority compared to the CM.

KEYWORDS collagen matrix, collagen-based matrix, dental implants, gingival biotype, gingival thickening, mucoderm®, Porcine collagen matrix, SCTG, soft tissue augmentation, subepithelial connective tissue graft

146 | wileyonlinelibrary.com/journal/jcpe J Clin Periodontol. 2021;48:146–163. SCHMITT et al. | 147

1 | INTRODUCTION Clinical Relevance Hard and soft tissue alterations after tooth extraction lead to de- Scientific rationale for study: To quantify soft tissue volume fects in the alveolar ridge, which becomes predominantly prominent alterations in the early healing phase after soft tissue thick- at the buccal contour (Araujo et al., 2015; Buser, et al., 2013; Buser, ening around implants with a native collagen matrix (CM) et al., 2013). This is especially of concern in the aesthetic zone, when in comparison to the autologous subepithelial connective planning an implant treatment (Araujo et al., 2015; Buser et al., 2011). tissue graft (SCTG). Several treatment concepts have been initiated to overcome such Principal findings: The healing after soft tissue thickening tissue deficiencies by means of bone and/or soft tissue augmenta- goes along with a large volume and thickness loss in the tion procedures. Some of the most common treatments are guided first three months (more pronounced, statistically insig- bone regeneration (GBR) procedures for buccal contour augmen- nificant, in the CM group). Between 3 and 6 months, soft tations, mostly combined with early or delayed implant placement tissue volume and thickness stabilizes. (Buser et al., 2011; Chappuis et al., 2017). After the healing period, it Practical implications: The tested collagen matrix could is apparent in many cases that the buccal contour is nevertheless not serve as an alternative to the SCTG for soft tissue thick- completely comparable with the surrounding regions, and a buccal ening. Clinicians need to assume significant volumetric tissue deficit still predominates (Zeltner et al., 2017). changes in the early healing phase. To compensate for this fact, a soft tissue thickening procedure at the buccal aspect is an appropriate treatment concept. The subepithelial connective tissue graft (SCTG) is still regarded as the gold standard in clinical practice (Akcali et al., 2015; Jung et al., 2004; Schneider et al., a significant volume loss in the early healing phase (first month) 2011; Schwarz et al., 2014; Sculean et al., 2015; Thoma et al., 2014; (Schmitt, Matta, et al., 2016). After 3 months of healing, the tissue Zuhr et al., 2014). However, harvesting procedures are associated with thickness finally stabilizes up to 10 months. Additionally, histological additional operative risks and postoperative patient morbidities and and immunohistological outcomes after 10 months of healing show a graft size is limited. Alternatives to autologous soft tissue grafts have comparable quality of the augmented connective tissues for the CM been a focus for a few years. For example, collagen-based matrices are and the SCTG (Schmitt et al., 2019). Thus far, there is not one clinical already being used clinically (Herford et al., 2010; Schmitt, Moest, et al., study that has used this specific collagen matrix for soft tissue thick- 2016; Schmitt et al., 2013; Thoma et al., 2016; Zeltner et al., 2017). ening in the current literature. Therefore, the present clinical study Their use seems promising since they eliminate the need for autologous aims to quantify 3D tissue alterations in the early healing phase (up transplantations (Bassetti et al., 2016; Ghanaati et al., 2011; Nocini et al., to 6 months) of the porcine collagen matrix (mucoderm®) versus the 2014; Sanz et al., 2009; Schmitt et al., 2013). For the regeneration of ke- SCTG for soft tissue thickening around dental implants. ratinized mucosa in open healing situations, such matrices have already been successfully used (Herford et al., 2010; Nocini et al., 2014; Schmitt et al., 2013). Although they do not perform as well as autologous free 2 | MATERIALS AND METHODS gingival grafts in terms of shrinkage of the augmented zone, their application seems justified in some situations (Schmitt et al., 2013). 2.1 | Study design For soft tissue thickening, various collagen matrices have also preclinically and clinically been tested (Schmitt, Matta, et al., 2016; This study was designed as a controlled clinical trial with a follow-up Thoma et al., 2011, 2016; Zeltner et al., 2017). In addition to good of 6 months (early healing phase). The aim of the current research tissue compatibility and nonirritating integration, a certain volume was to evaluate the natural porcine 3D collagen matrix (CM, muco- stability of the matrix was discussed as essential so that the matrix derm®) for soft tissue thickening around implants and to compare it can act as a placeholder for cell integration and therefore facilitate to the subepithelial connective tissue graft (SCTG). The sample size new tissue formation (Schmitt, Matta, et al., 2016; Zeltner et al., consideration was based on 3D outcomes after soft tissue thicken- 2017). In a former preclinical study of our working group, we al- ing with the SCTG published by Zeltner et al. (2017). Concerning the ready investigated the collagen matrix mucoderm® (CM, botiss bio- increase in the mucosal thickness (mm) after 6 months, an equiva- materials GmbH, Zossen, Germany) as an alternative to the SCTG. lence test of means using two groups of seven patients achieved a Mucoderm® is an acellular three-dimensional collagen matrix that is power of 0.94 at a significance level of 0.05 when the means were derived from the porcine dermis. It consists of type I and III collagen truly equivalent. The gingival thickness increase was assumed to be and has a natural structure (Ramachandra et al., 2014). The matrix 1.00 mm ±0.3 mm, and the equivalence margin was set to 0.4 mm. serves as a scaffold for ingrowing blood vessels and cells (Pabst et al., The calculated sample size is in accordance with comparable clinical 2014) and is completely degraded and replaced by new connective data published by Zeltner et al., 2017. They also calculated a number tissue within approximately 6–9 months (Rothamel et al., 2014). of seven patients per group (total 14 patients) with a power of 93%, Data from our former preclinical study has shown, that the in- assuming a standard deviation per group of 0.5 mm (Zeltner et al., tegration process of the matrix and the SCTG is associated with 2017). 148 | SCHMITT et al.

Upon approval by the ethical committee of the medical faculties • Mean distance (MeanD) in mm in the augmented region. The of the Friedrich-Alexander-University Erlangen-Nuremberg (approval mean distance describes the arithmetic mean deviation (mm) of number 298_14B), patients with the need for buccal soft tissue vol- all points of the surface comparison of two objects. ume increases at implant sites were consecutively recruited. The fol- • Maximum distance (MaxD) in mm in the augmented region. The lowing inclusion criteria were applied: maximum distance describes the maximum deviation (mm) of the surface comparison of two objects. • Minimum distance (MinD) in mm in the augmented region. The 2.1.1 | Inclusion criteria minimum distance describes the minimum deviation of the surface comparison of two objects. 1. >18 years of age, no pregnancy or breast feeding 2. Systematically healthy 3. Periodontally healthy (no probing pocket depths >4 mm) 2.3 | Clinical procedure 4. Situation after early implant insertion and simultaneous contour augmentation (guided bone regeneration, GBR) at least 4 months 2.3.1 | Surgery and a maximum of 6 months prior to enrolment 5. Necessity for soft tissue volume augmentation at the buccal im- Thickening of the buccal soft tissues was performed at the time plant site of implant exposure. Prior to surgery, an impression (i1) was taken with a scannable Vinylsiloxanether impression material (Identium® Scan Medium, Kettenbach GmbH & Co KG, Eschenburg, Germany). 2.1.2 | Exclusion criteria After local anaesthesia, a crestal submucosal incision was performed, and a split-thickness flap was raised with sharp dissection 1. Smoking using micro-blades. Implants were exposed with a sharp incision to 2. Diabetes the bone, and the healing screws were removed. At the border be- 3. Patients during, after or prior planned bisphosphonate therapy tween the crest and the buccal aspect, the split-thickness flap was 4. Patients during, after or prior radiation therapy of the oral-maxil- extended, creating a buccal split-thickness pouch. The dimensions lofacial complex of the pouch included the buccal aspect of the implant site and the 5. Any chemotherapy within the last 4 years buccal aspects of the two adjacent teeth. Then, a healing abutment 6. Any metabolic diseases that would negatively affect soft tissue that was approximately two mm higher than the expected flap healing thickness was screwed to the uncovered implant (Figure 1, c left 7. Previous or concurrent medication affecting wound healing, such and right). In the CM group, the matrix was rehydrated with sterile as steroids or any other kinds of anti-inflammatory drugs saline solution and then trimmed with scissors to fit into the buccal pouch (Figure 1, b right). The matrix, was used in one layer with Following inclusion in the study solely by the first Author (CMS), a thickness ranging from 1.2 to 1.7 mm (prior rehydrating). The cr- patients were scheduled for a screening visit. At this time point, estal aspect of the split-thickness flap and the underlying CM was study-related consent forms were completed, and patients were alter- then sutured to the palatal flap with interrupted sutures (resorb- nately assigned to one or the other treatment group. able, Vicryl 6.0, Ethicon GmbH & Co KG, Norderstedt, Germany) at the mesial and distal aspects of the healing screw (Figure 1, c right). In the SCTG group, the transplant was harvested from 2.2 | Outcome variables the palate in the region from the canine to the 2nd premolar/1st molar. After local anaesthesia of the palate, a paramarginal palatal 2.2.1 | Primary outcome variable subepithelial incision was made, and a subepithelial split-flap was sharply dissected towards the midline of the palate. The SCTG was • Integrated distance (ID) in mm3 (volume increase) in the aug- harvested with the periosteum, and the wound bed was imme- mented region (buccal implant site). The Integrated distance de- diately treated with local haemostatic measures. The SCTG was scribes the average volume increase from the arithmetic mean apically fixed with one mesial and one distal matrass suture (Vicryl of distances of the two surfaces to be compared to each other in 6.0). At the coronal aspect, the SCTG was sutured in the same way mm3. as the CM (Figure 1, c left). After surgery, an additional impression (i2) was taken. Postoperatively, analgesics (600 mg ibuprofen, HEXAL AG, Holzkirchen, Germany) and patients were instructed 2.2.2 | Secondary outcome variables to rinse with a 0.2% solution of chlorhexidine (GlaxoSmithKline Consumer Healthcare GmbH & Co. KG, Munich, Germany) twice a • Integrated distance (ID%, volume increase in%) in % (postoperative day for 5 days and to practice proper oral hygiene, with particular integrated distance defined as 100%) in the augmented region caution in the augmented area. SCHMITT et al. | 149

(a) (a)

(b) (b)

(d) (d)

(e) (e)

(f) (f) 150 | SCHMITT et al.

FIGURE 1 Exemplary clinical case presentation for the control (SCTG, left row a–f) and the test groups (CM, right row a–f). (a) Shows the situation prior to implant exposure and tissue thickening for both cases. Note the persisting tissue deficiency at the buccal contour. (b) (left) shows the harvested SCTG, and (b) (right) shows the rehydrated and trimmed CM. The substituted soft tissues were both inserted into a buccal pouch at the region of the implant with a split-thickness flap (c left and right) and then sutured with interrupted resorbable sutures (d left and right). E shows the time point of suture removal after ten days. In the test group (e, left), uneventful integration and wound healing were present in all cases. In the CM group (e, right), soft tissue swelling and redness were observed in most cases. (f) Shows the clinical situation at the 6-month examination time point for both groups

2.3.2 | Follow-up 2.5 | Statistical analyses

Control visits were performed one day, 10 days, 30 days, one Data were analysed with SPSS version 23.0 for Windows (IBM month and 6 months postoperatively. The augmented region was SPSS Software, Ehningen, Germany) and the statistical software clinically examined to detect any kind of wound healing complica- environment R V3.5.2 (R Core Team, 2018). R: A language and en- tion. Impressions i3, i4, i5 and i6 were taken at 10 days, 30 days, vironment for statistical computing. R Foundation for Statistical 3 months and 6 months. Prosthetic reconstruction was performed Computing, Vienna, Austria. URL https://www.R-project.org). in the period between the one- and three-month follow-up visits in Descriptive statistics included the arithmetic mean, max, min and all patients. standard deviation. The ID (mm3), the ID%, the MaxD (mm), the MeanD (mm) MinD (mm) were compared within the group over time (always with the 2.4 | 3D measurements situation directly after surgery) and between the groups at different examination time points. Impressions were scanned with an industrial scanner (ATOS II For intragroup comparisons, several mixed regression mod- SO4, Gom mbH, Braunschweig, Germany). ATOS Professional els were estimated. The ID (mm3 and %), MaxD, MeanD and MinD software (Gom GmbH) was used to match the virtual models. served as dependent variables, and the time point was used as a pre- Virtual model matching was performed in two steps. First, the dictor variable with ‘postsurgery’ as a reference. All examined areas two models to be matched were overlayed with a 3-point align- (1–6) of the models were also estimated separately. ment (assignment of three prominent structures to each other). For intergroup comparison, we also applied mixed regression In the second step, a ‘local best fit’ (assignment via individual models. Therefore, the group variables (CM and SCTG) were used structures within the scan, e.g. occlusal reliefs of adjacent teeth) as independent variables in addition to the analysed region and time was performed improving the initial matching outcome. This ena- point. bles a highly precise overlay of the CAD models. A region of in- To account for multiple testing, the significance levels were cor- terest (ROI) was individually selected for each augmented site rected using the method by Benjamini-Hochberg. For within-group (Figure 2). The augmented region was detected by matching the comparisons, the corrected significance level was 0.028. For be- baseline model (i1) with the one obtained directly after surgery tween-group comparisons, the significance level was set to 0.01. (i2). With this procedure, we were able to define the margins of each graft and precisely set the ROIs in the augmented regions. The set ROIs were standardized and used for all further follow- 3 | RESULTS up measurements of the corresponding sites by comparing i1-i2, i1-i3, i1-i4, i1-i5 and i1-i6. Additionally, the defined ROIs were di- 3.1 | Data set and patient demographics vided into 6 subregions (1–6, Figure 2) to evaluate 3D changes in the coronal-mesial, coronal-middle and coronal-distal regions as In total, 14 patients fulfilled the inclusion criteria and were included well as the apical-mesial, apical-middle and apical-distal regions in the study (Table 1). Seven patients with a mean age of 50.33 were of the ROIs (Figure 2). The rationale of additionally dividing the allocated to the CM group (4 men and 3 women), and seven patients ROI in 6 subregions was to investigate the impact of the graft with a mean age of 42.33 were allocated to the SCTG group (5 men localization on volume and thickness alterations after buccal soft and 2 women). tissue contour augmentation. Methodologically, this is comparable to the study of the working group of Schwarz and coworkers assessing volumetric tissue changes following combined surgical 3.2 | Clinical examination and follow-up therapy of peri-implantitis (Galarraga-Vinueza et al., 2020). The following data in the ROIs as well as the six subregions of the Postoperative follow-up was uneventful in all of the SCTG patients. ROIs were measured and recorded: ID, ID%, MeanD, MaxD and In the CM group, in five of the seven patients, the clinical exami- MinD. nation showed slight mucosal tissue swelling at the buccal, coronal SCHMITT et al. | 151

FIGURE 2 Shows the augmented region (region of interest, ROI) directly after surgery (coloured area). The exact locations of the ROIs were individually defined by superimposing the postoperative situation with the situation prior to surgery (i1–i2). Therefore, the extent of the augmented region was visible. The borders of the ROIs were then defined as follows: 1. lateral: the middle axis of the adjacent two teeth, 2. coronal: the most coronal part under the later position of the implant suprastructure and 3. apical: the mucogingival junction. In terms of inserting a frenulum in the defined ROI, this region was left out to avoid incorrect measurements. After defining the ROIs, they were divided into 6 subregions (1–6): 1. apical-mesial, 2. coronal-mesial, 3. apical-middle, 4. coronal-middle, 5. apical-distal and 6. coronal-distal. Therefore, all subregions could also be compared between groups and within groups over time to better localize the soft tissue changes

TABLE 1 Patient demographics and location of the augmented regions for each group

Sex (n) Age Region

Female Male Mean Median SD 12 11 21 22 23 43

CM (n = 7) 3 4 50.33 48.2 13.5 / 1 3 1 1 1 SCTG (n = 7) 2 5 42.33 40.5 12.2 1 2 2 2 / /

Abbreviations: CM, Collagen Matrix; SCTG, Subepithelial Connective Tissue Graft; SD, Standard Deviation.

aspect after ten days of healing. This soft tissue reaction, however, The volume increase (ID, mm3) after 6 months was rapidly declined in the following course and did not affect soft tissue 19.56 ± 8.95 mm3 in the CM group and 61.75 ± 52.69 mm3 in the healing or clinical outcomes (Figure 1). SCTG group, with no statistically significant difference. Expressed as a percentage, the results indicate a volume loss of the initially augmented soft tissue volume of 81.76% in the CM group and 3.3 | Three-dimensional tissue alterations (Table 2) 56.39% in the SCTG group after 6 months. The mean soft tissue thickness increase (MeanD) of the buccal contour after 6 months After surgery, there was a mean soft tissue thickness increase in the was 0.30 ± 0.16 mm in the CM group and 0.80 ± 0.61 mm in the buccal contour of 1.62 ± 0.28 mm (MeanD) in the CM group and SCTG group. Outcomes of the 6 months measurements are dis- 1.60 ± 0.6 mm (MeanD) in the SCTG group, with no statistically sig- played in Figure 4 for all patients. Subgroup analyses (1–6) also re- nificant difference. After the 10-day follow-up, the augmented soft vealed significant volume and thickness decreases in both groups tissue volume and soft tissue thickness decreased significantly up after six months compared to the situation after surgery. The graft to the three-month time point in both groups (Table 2, Figure 3). localization did not seem to have an impact on volumetric alter- Between three months and six months, the soft tissue volume and ations, since measurements showed heterogenous outcomes in thickness remained relatively unchanged. both groups (Table 2). 152 | SCHMITT et al. (Continues) -Value within within -Value N. a. p groups N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. 9.21 5.36 1.54 1.70 1.93 1.78 1.91 0.86 0.57 6.55 2.42 6.96 2.17 −5.57 −5.63 53.49 12.86 −0.04 100 100 100 100 100 Min 100 100 5.25 4.44 4.76 4.86 1.46 0.57 0.68 3.36 2.95 8.68 8.68 67.26 49.35 71.64 31.81 56.89 60.51 209.52 100 100 100 100 100 Max 100 100 0 0 0 0 0 0 0 1.07 1.05 1.15 1.18 0.91 0.57 0.82 2.24 2.09 2.30 1.97 8.22 19.47 19.95 57.21 14.02 20.57 20.07 SD 1.34 4.39 0.57 3.57 3.40 3.84 3.51 2.75 2.08 19.93 14.45 26.34 22.73 −0.49 −0.57 38.73 30.94 100 100 100 100 100 100 100 Mean 138.93 SCTG -Value within within -Value N. a. N. a. p groups N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. 1.85 1.19 1.83 1.54 1.53 1.06 0.87 0.57 2.83 3.29 2.72 2.80 13.00 52.69 −4.88 12.16 −4.88 −0.24 100 100 Min 100 100 100 100 100 5.96 5.14 5.98 4.26 5.98 1.86 1.77 0.53 0.12 3.48 3.85 29.10 55.22 71.42 23.21 32.66 22.71 190.08 100 100 100 100 100 100 Max 100 ) 3 9.95 0 0 0 0 0 0 0 1.15 1.51 1.54 0.87 0.91 0.66 0.73 0.51 2.08 6.71 1.97 1.23 8.52 19.37 51.02 18.70 12.94 SD 4.17 1.19 4.23 0.71 3.14 3.61 2.75 2.21 2.94 8.93 10.66 13.08 20.25 33.40 −0.69 38.83 −0.87 100 100 100 100 100 100 123.67 CM Mean 100 Showing the outcomes of the three-dimensional analyses for the control (subepithelial connective tissue graft; SCTG) and the test groups (collagen matrix; CM) 6 5 4 3 4 1 2 5 6 1 1 2 Total 3 Total Total 2 3 3 2 4 1 5 Total 6 Integrated Distance (Volume Increase, mm Maximum distance (mm) Minimum distance (mm) Integrated distance (volume increase, %) i1–i2 TABLE 2 SCHMITT et al. | 153 (Continues) -Value within within -Value N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. p groups N. a. N. a. .943 <.001 .007 .745 .004 .631 .001 .508 .003 <.001 .001 .034 .049 .001 .101 1.16 0.29 0.61 0.37 0.84 0.65 5.62 0.85 0.56 1.10 3.44 1.52 59.41 11.69 11.90 11.69 70.64 25.77 53.55 92.07 54.07 −0.04 −0.92 40.01 136.42 Min 1.95 3.63 0.57 3.74 3.92 2.18 2.60 2.33 2.60 8.35 2.76 55.55 56.79 66.44 60.67 43.46 88.52 207.25 313.56 285.08 312.67 Max 458.58 230.99 208.73 222.78 1.14 0.55 0.97 0.71 0.63 0.88 0.60 0.80 0.75 0.43 2.00 57.52 51.26 16.90 15.58 18.26 18.12 25.61 20.67 56.02 50.05 SD 58.52 86.33 83.91 114.65 1.86 1.30 1.55 1.50 0.89 0.68 4.42 2.33 2.63 1.60 16.30 35.51 26.80 22.02 −0.18 34.40 40.88 139.03 176.81 232.85 125.71 183.88 130.46 Mean 188.66 SCTG 122.20 -Value within within -Value N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. N. a. p groups .191 .016 .025 .01 .433 .009 .002 .666 .018 .04 <.001 .005 .307 .972 .003 9.22 1.15 1.36 1.77 1.01 0.67 0.64 0.68 5.73 0.09 1.58 4.18 0.64 0.56 1.33 1.98 52.83 26.69 62.66 22.52 −0.12 34.55 40.47 38.06 Min 1.51 1.64 0.12 0.99 3.28 2.28 4.07 2.38 2.92 2.30 2.08 19.25 10.74 21.77 18.35 36.50 30.35 147.06 145.92 196.96 115.00 101.82 36.01 131.56 126.95 133.47 Max ) 3 7.22 0.43 0.90 0.39 0.51 0.32 0.28 0.35 4.60 0.48 0.08 0.43 0.73 6.12 8.518 11.74 10.50 52.85 35.88 26.52 36.84 33.54 32.13 SD 34.46 40.24 9.76 1.72 1.93 1.77 1.06 1.17 1.34 5.75 0.01 0.59 2.23 1.62 2.87 67.36 19.99 57.09 69.57 11.46 11.19 24.19 71.03 73.69 80.74 88.76 CM Mean 100.55 6 2 3 4 5 Total 1 4 5 6 Total 2 Total Total 1 3 4 5 6 1 2 6 3 4 5 Integrated distance (volume increase, mm Mean distance (mm) Maximum distance (mm) Integrated distance (volume increase, %) i1–i3 TABLE 2 (Continued)TABLE 154 | SCHMITT et al. (Continues) -Value within within -Value .102 <.001 .117 .001 .161 <.001 .07 .255 .006 .995 .015 .947 p groups .003 .501 .046 .106 .007 .065 .955 .151 .431 .137 .013 .255 .011 .436 7.58 1.38 1.52 1.16 1.17 1.07 0.79 0.27 0.96 0.43 0.78 0.63 0.93 0.55 1.74 0.16 0.17 0.71 0.26 1.07 2.34 8.17 27.80 −3.96 −4.97 −0.08 Min 5.91 5.53 5.66 5.18 5.65 4.88 4.72 4.84 4.22 4.06 4.11 3.16 3.43 3.94 3.72 8.35 2.69 1.07 2.81 2.45 37.03 51.00 70.26 25.16 30.43 255.09 Max 9.51 1.27 1.12 1.08 1.04 1.29 1.15 1.33 1.30 1.33 1.45 1.40 1.40 0.98 1.09 0.84 0.93 2.07 1.07 2.24 2.64 14.33 10.48 16.88 26.29 SD 92.55 9.89 1.81 4.20 1.24 3.08 3.80 3.59 3.84 3.08 3.53 0.88 0.83 2.64 2.31 0.85 2.76 2.27 1.07 2.05 2.33 2.31 11.11 24.27 21.82 18.44 Mean 108.80 SCTG -Value within within -Value .513 .005 .872 .018 .002 .029 .192 .509 .191 .009 .007 .246 .003 p groups .889 .98 .539 <.001 .035 .245 .046 .563 <.001 .001 .008 .199 1.43 1.17 1.41 1.98 1.47 5.57 0.91 0.53 0.45 0.48 0.27 0.38 0.65 0.76 0.11 0.20 0.10 0.20 0.04 0.29 0.88 .006 −2.87 −2.87 24.25 −0.89 −0.05 Min 7.56 1.45 4.07 4.39 1.36 1.26 1.97 1.06 0.48 3.32 3.10 0.21 0.52 2.66 2.36 2.11 2.12 2.74 2.87 1.86 2.25 11.32 11.05 15.92 85.58 33.15 Max ) 3 1.06 0.61 0.64 0.57 0.60 1.24 0.43 0.66 0.87 0.63 0.48 0.85 4.23 0.40 4.48 0.41 0.48 0.73 0.44 0.75 0.37 6.61 1.18 2.19 11.02 23.32 SD 7.40 1.62 1.58 1.04 1.20 1.01 0.79 4.32 1.00 0.01 0.91 0.40 0.31 2.77 2.28 2.90 2.03 2.14 2.17 3.61 1.12 6.08 16.20 50.66 −0.72 −0.76 CM Mean 3 4 5 6 2 2 3 4 5 6 1 1 1 Total Total Total 2 3 6 4 1 2 4 5 3 5 Integrated distance (volume increase, mm Mean distance (mm) Minimum distance (mm) i1–i4 TABLE 2 (Continued)TABLE SCHMITT et al. | 155 (Continues) -Value within within -Value .439 .027 .07 .023 .858 .986 .111 .556 p groups .997 .032 .303 .039 .052 .321 .3 .672 .988 .002 .179 .61 .584 .004 .03 .004 .879 .951 7.98 1.32 1.06 0.08 0.45 0.45 0.29 0.72 0.68 0.54 0.32 3.67 0.81 0.75 1.43 −1.12 −1.19 11.00 52.38 45.67 20.92 22.80 30.86 −0.21 −0.14 −0.01 Min 4.02 4.23 4.43 4.68 4.28 4.50 1.26 1.35 0.72 3.73 0.94 3.01 3.44 3.04 3.62 8.62 8.62 2.98 47.98 79.24 93.80 209.55 142.37 143.44 146.74 160.33 Max 1.32 1.32 1.41 1.25 1.36 1.63 1.29 1.42 1.39 1.54 0.94 0.72 0.58 0.81 0.61 2.70 1.01 2.65 17.30 49.83 74.20 41.13 21.81 26.29 43.38 46.01 SD 1.56 1.23 1.56 1.06 1.37 0.72 0.43 0.83 0.59 3.71 3.65 2.15 2.19 2.10 2.10 2.53 1.29 57.48 79.55 18.99 72.63 82.52 92.24 54.48 −0.09 116.59 Mean SCTG -Value within within -Value <.001 <.001 <.001 <.001 .001 .002 <.001 .001 <.001 <.001 <.001 p groups <.001 .681 .309 .087 <.001 .001 <.001 .041 .131 .018 <.001 .623 .357 .115 1.47 5.68 0.33 0.09 0.53 0.81 0.45 0.79 0.56 0.84 .004 0.92 0.98 19.82 −1.51 −1.51 24.17 13.06 15.75 18.34 23.24 −0.02 −0.44 −0.31 −0.31 −0.43 −0.28 Min 1.44 1.53 5.99 5.99 1.17 1.16 4.83 1.82 0.86 0.62 0.11 0.26 0.65 3.25 2.24 2.50 1.14 76.33 95.16 21.64 91.86 82.07 82.58 68.43 84.42 −0.05 Max 7.17 1.48 0.56 0.42 0.30 0.43 0.53 0.58 0.52 0.92 0.62 0.51 0.69 0.49 0.30 0.25 0.32 2.07 1.75 21.82 26.04 25.32 25.29 25.39 28.14 20.60 SD 1.86 1.46 1.63 1.72 1.02 0.72 0.86 0.42 0.10 0.34 0.35 0.69 0.10 2.26 2.77 45.67 45.66 13.54 42.91 33.80 43.80 38.69 43.99 −0.31 −0.12 −0.50 CM Mean 2 3 1 5 Total 6 4 6 2 3 1 1 1 2 3 6 4 5 2 3 Total 4 5 Total Total 6 Integrated distance (volume increase, %) Maximum distance (mm) Minimum distance (mm) Mean distance (mm) TABLE 2 (Continued)TABLE 156 | SCHMITT et al. (Continues) -Value within within -Value <.001 .197 .201 <.001 .717 .001 .026 <.001 p groups .01 .001 .263 <.001 .148 .003 .117 .057 .002 .699 .002 .157 .084 <.001 .001 .362 .005 1.27 1.15 0.88 1.10 0.73 0.36 0.13 0.60 5.50 0.50 0.33 3.91 0.26 1.37 2.2 6.42 6.26 27.16 39.59 29.58 41.47 11.00 10.19 20.92 38.06 Min 4.05 3.03 3.57 2.97 0.33 2.99 2.83 2.27 2.82 8.61 8.61 79.24 29.33 29.09 99.85 96.39 26.25 25.16 93.80 90.25 83.04 30.66 48.84 221.64 Max 156.89 9.33 9.51 1.17 1.34 0.79 0.88 0.65 0.94 0.95 0.82 0.33 2.65 8.51 8.58 2.48 19.82 11.04 21.50 45.12 21.81 15.47 26.29 70.22 SD 22.25 22.39 1.76 1.66 1.36 1.87 1.86 1.93 0.96 0.33 2.03 3.22 2.83 17.68 57.48 59.03 69.43 77.00 11.54 11.11 61.92 10.42 16.34 92.02 86.60 20.06 54.48 Mean SCTG -Value within within -Value <.001 <.001 <.001 <.001 .002 <.001 <.001 <.001 .014 p groups <.001 <.001 .502 .005 <.001 <.001 <.001 .001 .002 <.001 <.001 <.001 <.001 <.001 .005 <.001 0.67 0.26 0.45 0.46 0.17 0.17 0.35 0.40 0.11 0.86 5.41 0.95 0.59 0.35 −9.13 −2.12 −2.44 −5.67 10.50 10.64 −8.86 −0.76 −0.45 −11.51 −28.11 Min 9.49 1.11 1.82 1.40 1.43 1.50 1.58 1.14 5.57 1.46 0.99 0.13 1.82 6.91 8.48 8.36 47.38 19.14 39.66 75.68 45.7 32.83 83.96 60.28 40.18 Max ) 3 0.19 0.60 0.35 0.42 0.30 0.53 0.45 0.32 4.44 0.54 0.34 1.20 3.41 3.75 2.56 2.41 17.01 19.05 14.69 11.28 24.10 16.82 28.74 28.32 SD 34.30 1.00 0.91 0.90 0.74 0.88 0.49 1.82 0.94 0.79 4.16 0.83 2.61 1.35 3.80 6.18 6.41 19.82 39.14 −1.45 14.18 24.90 28.37 12.80 22.46 22.19 CM Mean 6 5 6 4 5 3 4 2 1 Total 1 Total Total 2 3 2 5 6 1 4 4 5 Total 3 6 Integrated distance (volume increase, mm Maximum distance (mm) Minimum distance (mm) Integrated distance (volume increase, %) i1–i5 TABLE 2 (Continued)TABLE SCHMITT et al. | 157 (Continues) -Value within within -Value .014 p groups .403 .337 .045 .242 .06 .008 .097 .001 .061 <.001 .765 .32 <.001 <.001 .168 .082 .062 .049 <.001 <.001 <.001 .014 .074 .106 <.001 7.99 0.35 0.11 0.05 0.45 0.23 0.13 0.64 0.61 1.7 2.73 −5.75 −5.99 −4.58 −6.79 28.80 −0.27 −0.41 −0.22 −0.67 −0.09 −17.56 −19.48 −15.40 −26.36 Min −104.78 1.36 1.28 1.95 1.87 1.20 0.51 2.01 2.17 2.21 2.46 2.56 2.46 1.85 27.02 94.14 21.12 26.98 85.16 23.71 25.17 72.55 98.57 92.56 44.20 121.36 Max 148.07 9.11 9.58 0.63 0.48 0.74 0.74 0.55 0.80 0.49 0.56 0.71 0.67 0.68 0.39 0.44 8.42 8.03 67.51 24.84 10.86 15.28 52.69 SD 34.13 22.49 40.27 30.02 7.81 7.48 9.63 1.13 1.00 1.02 1.45 1.26 1.53 0.78 0.64 0.96 0.36 0.48 1.10 47.96 61.70 14.48 61.75 10.08 12.77 −0.02 43.61 60.65 38.42 43.27 Mean SCTG -Value within within -Value 0 .001 .001 .003 0 .725 0 0 0 0 0 p groups .579 .034 <.001 .002 .013 <.001 <.001 <.001 .001 <.001 <.001 <.001 <.001 .001 <.001 9.98 0.20 0.04 1.45 0.04 0.12 0.40 8.02 −9.19 −1.16 −2.11 −5.67 −3.96 15.37 −4.68 12.06 −0.12 −0.03 −0.14 −0.46 −0.50 −0.40 −0.25 −17.41 −14.20 Min −34.68 7.94 9.90 1.01 5.76 0.61 0.13 0.21 0.62 0.57 0.76 0.78 0.62 0.44 0.16 0.67 0.20 6.50 8.99 49.66 79.78 11.59 35.49 42.05 23.88 78.15 34.25 Max ) 3 1.86 5.09 0.17 1.40 0.30 0.27 0.43 0.50 0.27 0.67 0.28 0.25 0.22 2.47 2.10 3.56 3.81 2.56 8.95 8.72 14.48 10.45 21.71 42.08 18.54 22.73 SD 4.47 0.36 0.22 0.28 0.48 0.35 0.35 0.32 3.34 3.45 2.54 2.95 2.87 8.32 19.56 −1.29 39.12 29.38 10.72 18.24 23.86 −0.93 −0.72 −0.13 −0.08 −0.12 CM Mean 1 6 2 3 4 5 1 2 3 4 5 Total 6 Total 5 1 6 Total 2 3 4 1 5 2 4 3 Integrated distance (volume increase, mm Mean distance (mm) Integrated distance (volume increase, %) i1–i6 TABLE 2 (Continued)TABLE 158 | SCHMITT et al. -Value within within -Value <.001 .046 p groups <.001 .001 <.001 <.001 .003 <.001 .678 .956 .008 .001 .525 .179 .038 <.001 .005 <.001 .055 <.001 .212 .041 0.14 0.14 0.68 0.12 0.68 0.58 0.41 0.49 0.15 0.07 ): .058; integrated distance (%): −1.08 −1.16 −2.04 −5.29 −5.29 −0.47 −0.34 −0.36 −0.27 −0.36 −0.16 3 −27.71 Min 1.75 1.87 1.73 1.32 1.64 0.49 0.94 3.36 2.64 2.54 2.61 0.49 2.58 0.71 8.58 8.58 2.12 2.34 2.30 2.19 1.74 -value comparison within groups always means 80.78 Max 0.87 1.97 1.19 0.87 0.78 0.91 0.72 0.49 0.69 1.87 0.71 0.60 0.98 0.78 0.82 0.61 0.55 0.64 2.55 0.58 2.52 months after surgery (i6) were compared with the situation prior to 37.18 SD

1.29 1.28 1.49 1.72 1.44 1.04 1.04 1.06 0.49 0.63 0.07 0.73 0.54 0.56 0.80 0.84 2.49 0.10 2.70 39.36 −0.53 −0.86 Mean SCTG months after surgery (i5) and 6

-values for the intragroup comparisons are given. The p -values for intergroup comparisons were as follows: integrated distance (mm -Value within within -Value <.001 <.001 <.001 <.001 <.001 .095 p groups <.001 .308 .087 <.001 .582 .001 <.001 <.001 <.001 <.001 <.001 .399 <.001 .001 0.44 0.51 0.22 0.43 0.48 0.28 .001 0.03 0.89 0.01 0.20 0.17 0.14 −1.04 −2.39 −2.55 month after surgery (i4), 3 −5.62 −4.36 −0.27 .027 −0.68 −0.20 −0.03 −0.28

Min ) for each parameter and p SD 5.62 1.21 1.94 0.96 1.00 0.13 0.53 0.25 0.52 0.31 0.56 0.82 0.79 0.60 0.08 2.00 0.55 2.80 2.85 46.88 −0.01 −0.17 Max days after surgery (i3), 1

1.98 0.20 0.55 0.28 1.01 0.87 0.49 0.30 0.69 0.42 1.00 0.17 0.46 0.24 0.41 0.25 0.31 0.14 0.16 2.61 1.74 15.90 SD 1.01 1.72 1.19 1.13 0.70 0.79 0.06 0.26 0.22 0.38 0.26 0.21 0.54 0.30 1.50 −1.66 −1.28 18.17 −0.35 −0.37 −0.11 −0.51 CM Mean 6 6 4 5 1 2 3 1 2 3 Total Total 4 6 4 5 5 1 2 3 6 Total Maximum distance (mm) Minimum distance (mm) Mean distance (mm) surgery (i1). The mean, maximum, minimum and standard deviation ( the comparison with the situation directly after soft tissue augmentation The global (i1–i2). p Note: The situations directly after surgery (i2), 10 .014; maximum.014; distance (mm): .093; minimum distance and (mm): mean .173; distance (mm): .071. Since there were no statistically significant differencesapplied. The detected, significance single level group for intragroup comparisons comparisons were not was .028 and for intergroup comparisons was .01. Bold numbers highlight the outcomes for the total region of interest. TABLE 2 (Continued)TABLE SCHMITT et al. | 159

(a) (a)

(b) (b)

(d) (d)

(e) (e)

FIGURE 3 Exemplary presentation of superimposed 3D data sets for the control (SCTG left row a–e) and the test groups (CM right row a–e). (a) Shows the situation directly after soft tissue augmentation (i1–i2). Note the volume increase at the buccal contour. B shows the situation after ten days of healing (i1–i3). (c) shows the situation after one month (i1–i4). (d) shows the situation after three months (i1–i5). (e) shows the situation after six months (i1–i6). (d and e) Show the situation with the prosthetic restoration that was performed between the one- and three-month follow-up periods in all cases. In both groups, the augmented soft tissue volume significantly decreased over time (a–e) 160 | SCHMITT et al.

FIGURE 4 Showing the three- dimensional outcomes after 6 months of healing for the SCTG (left row) and for the CM groups (right row)

4 | DISCUSSION

In this study, the collagen matrix mucoderm® (CM) was evaluated as an alternative soft tissue graft to the SCTG focusing on three- dimensional changes in the buccal contour with a 3D follow-up of the early healing phase. In both groups, the soft tissue grafts (CM and SCTG) showed sufficient clinical integration with no signs of inflammation. The outcomes of the 3D measurements revealed significant volume and thickness decreases in the early healing phase (3 months) after soft tissue thickening in both groups. Between 3 and 6 months, only minimal changes were observed. In preclinical models, the matrix shows an acceptable tissue response with almost no signs of ingrowing inflammatory cells and a resorption time of 8–12 weeks (Rothamel et al., 2014). In the context of matrix degradation, the formation of new connective tissue can be histologically observed, and the matrix is partially replaced by new connective tissue. This process, however, seems to be associated with volumetric changes, and the dimension of the initially inserted matrix does not correlate with the gained new connective tissue volume, as shown in this study and a former preclinical 3D study of our working group. The outcomes of these measurements revealed that only 27.08% in the SCTG group and 11.03% in the CM group of the former augmented soft tissue volume was left after 10 months of healing (Schmitt, Matta, et al., 2016). In the present clinical study, the same 3D measuring protocol was applied. The methodology has already been shown to be very precise for detecting subtle 3D changes. In this study, soft tissue volume changes were not as pronounced as they were in the preclinical study. In the SCTG group, the volume loss of the originally augmented region was 72.92% in the preclinical study (10-month follow-up) and only 56.39% in the clinical study (6-month follow-up). This may be attributed to the different clinical scenarios or to the different examination time points (10 and 6 months). However, in the CM group, the outcomes showed similar results when comparing the preclinical and clinical studies with each other (89.97% and 81.76% volume loss after 10 and 6 months). We assume that the tremendous volume loss in the CM groups is attributed to the fast matrix degradation and tissue turnover that takes place in the first 3 months after soft tissue thickening (Schmitt, Matta, et al., 2016), leading to an inferior result of the CM compared to the SCTG. After the early healing phase (3 months), tissue volumes and thicknesses stabilized in both groups with marginal changes up to 6 or 10 months. According to the results of the preclinical study and the outcomes of the present study, it can be assumed that no significant changes will result after 6 months. This has also been confirmed SCHMITT et al. | 161 by histological and immunohistological connective tissue character- split-thickness flap. The VCMX or the SCTG were placed on top of ization after soft tissue thickening around teeth with either the CM the ridge and underneath the buccal flap, with primary wound clo- or the SCTG. Descriptive histological analyses showed no residual sure. The soft tissue thickness increases were 1.4 ± 1.4 mm (occlusal), soft tissue grafts and a mature connective tissue with no differences 1.1 ± 1.4 mm (buccal) and 0.9 ± 1.9 (apical) after 90 days in the VCMX between the groups (Schmitt et al., 2019). group and 0.8 ± 1.8 (occlusal), 0.8 ± 2.2 mm (buccal) and 1.6 ± 2.6 In a recently published clinical pilot study (Papi & Pompa, 2018), (apical) in the SCTG group (no statistically significant differences). the CM (mucoderm®) was used to improve the keratinized mucosa They (Zeltner et al., 2017) also published linear volumetric soft width (KMW, mm) around the inserted implants. The CM was in- tissue changes up to 3 months in the same patient cohort. In the serted under a buccal full-thickness flap in the context of implant ex- VCMX group, the crestal change was 0.27 ± 0.26 mm, and the buccal posure and healing abutment connection. No membrane exposures change was 0.77 ± 0.74 mm; in the SCTG group, the changes were or wound healing complications occurred during the postoperative 0.42 ± 0.74 (crestal) and 0.79 ± 0.45 mm (buccal) after 90 days. phase, and the KMW was 7.86 ± 3.22 mm after one month (100%) The therapeutic approaches are quite comparable to those used in and 5.67 ± 2.12 mm (72.13%) after one year. The surgical approach is our study (Thoma et al., 2016; Zeltner et al., 2017). However, there are comparable to that of our study, except that we inserted the CM into few small differences. 1. Thoma and coworkers performed soft tissue a split-thickness flap. Although a soft tissue thickening procedure thickening with an additional surgery prior to implant exposure (coro- was performed in the study by Papi & Pompa, thickness alterations nal and buccal aspects). We augmented the soft tissues (buccal aspect) were not measured. However, the results of their study showed at the time of implant exposure. 2. We inserted the soft tissue graft in uneventful healing in all cases (n = 12). As already mentioned, we a split-thickness flap. 3. The measuring process used for the soft tissue observed slight soft tissue swelling and redness at the buccal cor- changes was different (Thoma et al., 2016; Zeltner et al., 2017). They onal contour in few cases on the tenth postoperative day. It can be measured the thickness gain with a two-dimensional method (trans- assumed that the positioning of the soft tissue graft has an influence mucosal probing) and a 3D method calculating the volumetric changes on the biological response and therefore on the healing outcome. measured in mm, which corresponded to the mean distance between We also know from a preclinical study that used a cross-linked the three surfaces between the different time points. The 3D method collagen matrix for tissue thickening in chronical ridge defects that can be compared to the 3D method applied in our study used to mea- subperiosteal placement of collagen matrices results in bone and sure the thickness gain (MeanD) in mm in the augmented region. soft tissue increases (Thoma et al., 2011). Although this finding has The thickness gain of the buccal soft tissues in the SCTG group not been histologically confirmed, it is hypothesized that insertion (MeanD) in our study (1.10 ± 0.49 mm in the augmented region) after of the CM in a split-thickness flap (as was done in this study) only 90 days was comparable to the 2D and 3D measurements of Thoma et al. results in soft tissue changes, and the underlying bone remains un- at the buccal site of the ROI (Thoma et al., 2016; Zeltner et al., 2017). The changed. For the long-term stability and health of peri-implant tis- thickness increase (MeanD) with the native porcine collagen matrix used sues, it can be discussed whether soft and bone tissue increases by in our study (0.32 ± 0.25 mm) was inferior to the outcomes obtained with subperiosteal placement or only soft tissue increases by epiperios- the VCMX at the buccal sites of the ROI (Thoma et al., 2016). teal placement of collagen matrices should be demanded. With respect to the volumetric soft tissue changes additionally Additional studies that use the collagen matrix used in our study obtained with the 3D measuring method applied in our study (inte- do not yet exist in the current literature. However, soft tissue thick- grated distance in mm3 and %), there was no comparability with the ening procedures with a volume-stable cross-linked porcine colla- other data in the current literature. gen matrix (VCMX, Fibro-Gide, Geistlich Pharma AG, Wolhusen. Examining the available data also makes it clear that in addition Switzerland) have been investigated in multiple preclinical and one to the collagen matrix used, the following factors can certainly affect clinical scenario (Thoma et al., 2011, 2017; Zeltner et al., 2017). The the outcomes and make the studies difficult to compare: 1. the in- approach is to prevent fast degradation by the cross-linking of col- dications, time points and surgical procedure itself; 2. the follow-up lagen-based matrices and to enhance connective tissue formation time points and the length of the follow-up; and finally 3. the meth- compared to non-cross-linked collagen matrices, as used in our odology of measuring the thickness or volume of the augmented soft study (Rothamel et al., 2014; Thoma et al., 2012). As a negative side tissues. Finally, the results of the studies already mentioned, includ- effect, cross-linking can favour inflammatory and foreign body reac- ing those of our study, should be interpreted with caution as long as tions, can lead to wound healing complications and can affect clinical there are no additional comparable studies with results that reflect outcomes (Annen et al., 2011; Becker et al., 2009). The biological a clear trend. response to VCMX showed a slight to moderate infiltration with in- Some limitations exist with respect to the present clinical study. flammatory cells during the first 2 months but not after 6 months First, a relatively small number of patients were treated, which may (confirmed by histology) (Thoma et al., 2017). Importantly, this had have influenced the outcome of the statistical comparisons between no significant effect on the healing outcome (Thoma et al., 2016). the groups. Additionally, the measuring method may have falsified the In the clinical study (20 patients), either the VCMX or SCTG was results due to the inaccuracies of the impressions taken. In the future, used to thicken the soft tissues at implant sites. This procedure was this limitation should be overcome by the implementation of digital performed with a full-thickness flap on top of the ridge and a buccal impression methods. 162 | SCHMITT et al.

5 | CONCLUSION Dental Research, 92, 176S–182S. https://doi.org/10.1177/00220 34513504949 Buser, D., Wittneben, J., Bornstein, M. M., Grutter, L., Chappuis, V., & Within the limitations of the study, 3D outcomes indicate that the Belser, U. C. (2011). Stability of contour augmentation and esthetic porcine collagen matrix used in this study is inferior (not statistically outcomes of implant-supported single crowns in the esthetic zone: significant) for soft tissue thickening procedures around implants 3-year results of a prospective study with early implant placement postextraction. , , 342–349. https://doi. than the gold standard of care, the SCTG. Journal of Periodontology 82 org/10.1902/jop.2010.100408 Further research should focus on the soft tissue substitute Chappuis, V., Rahman, L., Buser, R., Janner, S. F. M., Belser, U. C., & Buser, composition and the associated biological response (substitute in- D. (2017). Effectiveness of contour augmentation with guided tegration, degradation and new tissue formation) and the impact bone regeneration: 10-Year results. Journal of Dental Research, of the placement of the substitute (full-thickness or split-thickness 97(3), 266–274. https://doi.org/10.1177/0022034517737755 Galarraga-Vinueza, M. E., Obreja, K., Magini, R., Sculean, A., Sader, R., & approach). Schwarz, F. (2020). Volumetric assessment of tissue changes following combined surgical therapy of peri-implantitis: A pilot study. ACKNOWLEDGEMENT Journal of Clinical Periodontology, 47(9), 1159–1168. https://doi. The present work was part of the studies of the second author org/10.1111/jcpe.13335 Ghanaati, S., Schlee, M., Webber, M. J., Willershausen, I., Barbeck, M., Brückbauer P to obtain a doctoral degree (Dr. med. dent.) at the Balic, E., Gorlach, C., Stupp, S. I., Sader, R. A., & Kirkpatrick, C. J. University of Erlangen- Nuremberg, Erlangen, Germany. Open access (2011). Evaluation of the tissue reaction to a new bilayered collagen funding enabled and organized by Projekt DEAL. matrix in vivo and its translation to the clinic. Biomedical Materials, 6, 15010. https://doi.org/10.1088/1748-6041/6/1/015010 Herford, A. S., Akin, L., Cicciu, M., Maiorana, C., & Boyne, P. J. (2010). CONFLICT OF INTEREST Use of a porcine collagen matrix as an alternative to autogenous The authors declare no conflicts of interest related to this study. tissue for grafting oral soft tissue defects. Journal of Oral and Maxillofacial Surgery, 68, 1463–1470. https://doi.org/10.1016/j. ORCID joms.2010.02.054 Jung, R. E., Siegenthaler, D. W., & Hammerle, C. H. (2004). Postextraction Christian M. Schmitt https://orcid.org/0000-0002-9028-5139 tissue management: a soft tissue punch technique. The International Journal of Periodontics & Restorative Dentistry, 24, 545–553. REFERENCES Nocini, P. F., Castellani, R., Zanotti, G., Gelpi, F., Covani, U., Marconcini, Akcali, A., Schneider, D., Unlu, F., Bicakci, N., Kose, T., & Hammerle, S., & de Santis, D. (2014). Extensive keratinized tissue augmentation C. H. (2015). Soft tissue augmentation of ridge defects in the during implant rehabilitation after Le Fort I osteotomy: using a new maxillary anterior area using two different methods: a random- porcine collagen membrane (Mucoderm). Journal of Craniofacial ized controlled clinical trial. Clinical Oral Implants Research Clinical Surgery, 25, 799–803. https://doi.org/10.1097/SCS.0000000000 Oral Implants Research, 26, 688–695. https://doi.org/10.1111/ 000524 clr.12368 Pabst, A. M., Happe, A., Callaway, A., Ziebart, T., Stratul, S. I., Ackermann, Annen, B. M., Ramel, C. F., Hammerle, C. H., & Jung, R. E. (2011). Use M., Konerding, M. A., Willershausen, B., & Kasaj, A. (2014). In vitro of a new cross-linked collagen membrane for the treatment of and in vivo characterization of porcine acellular dermal matrix for peri-implant dehiscence defects: a randomised controlled dou- gingival augmentation procedures. Journal of Periodontal Research, ble-blinded clinical trial. European Journal of Oral Implantology, 4, 49, 371–381. https://doi.org/10.1111/jre.12115 87–10 0. Papi, P., & Pompa, G. (2018). The use of a novel porcine derived acel- Araujo, M. G., Silva, C. O., Misawa, M., & Sukekava, F. (2015). Alveolar lular dermal matrix (Mucoderm) in peri-implant soft tissue aug- socket healing: what can we learn?. Periodontology 2000, 68, 122– mentation: Preliminary results of a prospective Pilot Cohort 134. https://doi.org/10.1111/prd.12082 Study. BioMed Research International, 2018, 6406051. https://doi. Bassetti, R. G., Stahli, A., Bassetti, M. A., & Sculean, A. (2016). Soft tissue org/10.1155/2018/6406051 augmentation procedures at second-stage surgery: A systematic Ramachandra, S. S., Rana, R., Reetika, S., & Jithendra, K. D. (2014). review. Clinical Oral Investigations, 20(7), 1369–1387. https://doi. Options to avoid the second surgical site: a review of literature. Cell org/10.1007/s00784-016-1815-2 and Tissue Banking, 15, 297–305. https://doi.org/10.1007/s1056 Becker, J., Al-Nawas, B., Klein, M. O., Schliephake, H., Terheyden, H., & 1-013-9395-8 Schwarz, F. (2009). Use of a new cross-linked collagen membrane R Core Team (2018). R: A language and environment for statistical com- for the treatment of dehiscence-type defects at titanium implants: puting. Vienna, Austria: R Foundation for Statistical Computing. A prospective, randomized-controlled double-blinded clinical mul- Available online at https://www.R-project.org/ ticenter study. Clinical Oral Implants Research, 20, 742–749. https:// Rothamel, D., Benner, M., Fienitz, T., Happe, A., Kreppel, M., Nickenig, doi.org/10.1111/j.1600-0501.2008.01689.x H. J., & Zoller, J. E. (2014). Biodegradation pattern and tissue inte- Buser, D., Chappuis, V., Bornstein, M. M., Wittneben, J. G., Frei, M., & gration of native and cross-linked porcine collagen soft tissue aug- Belser, U. C. (2013). Long-term stability of contour augmentation mentation matrices - An experimental study in the rat. Head & Face with early implant placement following single tooth extraction in Medicine, 10, 10. https://doi.org/10.1186/1746-160X-10-10 the esthetic zone: A prospective, cross-sectional study in 41 pa- Sanz, M., Lorenzo, R., Aranda, J. J., Martin, C., & Orsini, M. (2009). tients with a 5- to 9-year follow-up. Journal of Periodontology, 84, Clinical evaluation of a new collagen matrix (Mucograft proto- 1517–1527. https://doi.org/10.1902/jop.2013.120635 type) to enhance the width of keratinized tissue in patients with Buser, D., Chappuis, V., Kuchler, U., Bornstein, M. M., Wittneben, J. G., fixed prosthetic restorations: a randomized prospective clinical Buser, R., Cavusoglu, Y., & Belser, U. C. (2013). Long-term stability trial. Journal of Clinical Periodontology, 36, 868–876. https://doi. of early implant placement with contour augmentation. Journal of org/10.1111/j.1600-051X.2009.01460.x SCHMITT et al. | 163

Schmitt, C. M., Matta, R. E., Moest, T., Humann, J., Gammel, L., Neukam, (2011). Soft tissue volume augmentation by the use of colla- F. W., & Schlegel, K. A. (2016). Soft tissue volume alterations after gen-based matrices in the dog mandible – A histological analy- connective tissue grafting at teeth: the subepithelial autologous sis. Journal of Clinical Periodontology, 38, 1063–1070. https://doi. connective tissue graft versus a porcine collagen matrix - a pre-clin- org/10.1111/j.1600-051X.2011.01786.x ical volumetric analysis. Journal of Clinical Periodontology, 43, 609– Thoma, D. S., Naenni, N., Benic, G. I., Hammerle, C. H., & Jung, R. E. (2017). 617. https://doi.org/10.1111/jcpe.12547 Soft tissue volume augmentation at dental implant sites using a vol- Schmitt, C. M., Moest, T., Lutz, R., Wehrhan, F., Neukam, F. W., & Schlegel, ume stable three-dimensional collagen matrix – Histological out- K. A. (2016). Long-term outcomes after vestibuloplasty with a comes of a preclinical study. Journal of Clinical Periodontology, 44, porcine collagen matrix (Mucograft(R)) versus the free gingival 185–194. https://doi.org/10.1111/jcpe.12635 graft: a comparative prospective clinical trial. Clinical Oral Implants Thoma, D. S., Villar, C. C., Cochran, D. L., Hammerle, C. H., & Jung, R. Research, 27, e125–e133. https://doi.org/10.1111/clr.12575 E. (2012). Tissue integration of collagen-based matrices: An exper- Schmitt, C. M., Schlegel, K. A., Gammel, L., & Moest, T. (2019). imental study in mice. Clinical Oral Implants Research, 23, 1333– Gingiva thickening with a porcine collagen matrix in a pre- 1339. https://doi.org/10.1111/j.1600-0501.2011.02356.x clinical dog model: Histological outcomes. Journal of Clinical Thoma, D. S., Zeltner, M., Hilbe, M., Hammerle, C. H., Husler, J., & Jung, Periodontology, 46(12), 1273–1281. https://doi.org/10.1111/ R. E. (2016). Randomized controlled clinical study evaluating effec- jcpe.13196 tiveness and safety of a volume-stable collagen matrix compared Schmitt, C. M., Tudor, C., Kiener, K., Wehrhan, F., Schmitt, J., Eitner, S., to autogenous connective tissue grafts for soft tissue augmenta- Agaimy, A., & Schlegel, K. A. (2013). Vestibuloplasty: Porcine colla- tion at implant sites. Journal of Clinical Periodontology, 43, 874–885. gen matrix versus free gingival graft: a clinical and histologic study. https://doi.org/10.1111/jcpe.12588 Journal of Periodontology, 84, 914–923. https://doi.org/10.1902/ Zeltner, M., Jung, R. E., Hammerle, C. H., Husler, J., & Thoma, D. S. (2017). jop.2012.120084 Randomized controlled clinical study comparing a volume-stable Schneider, D., Grunder, U., Ender, A., Hammerle, C. H., & Jung, R. E. collagen matrix to autogenous connective tissue grafts for soft (2011). Volume gain and stability of peri-implant tissue following tissue augmentation at implant sites: Linear volumetric soft tissue bone and soft tissue augmentation: 1-year results from a prospec- changes up to 3 months. Journal of Clinical Periodontology, 44, 446– tive cohort study. Clinical Oral Implants Research, 22, 28–37. https:// 453. https://doi.org/10.1111/jcpe.12697 doi.org/10.1111/j.1600-0501.2010.01987.x Zuhr, O., Baumer, D., & Hurzeler, M. (2014). The addition of soft tissue re- Schwarz, F., Mihatovic, I., Shirakata, Y., Becker, J., Bosshardt, D., & placement grafts in plastic periodontal and implant surgery: critical Sculean, A. (2014). Treatment of soft tissue recessions at titanium elements in design and execution. Journal of Clinical Periodontology, implants using a resorbable collagen matrix: A pilot study. Clinical 41(Suppl. 15), S123–S142. https://doi.org/10.1111/jcpe.12185 Oral Implants Research, 25, 110–115. https://doi.org/10.1111/ clr.12042 Sculean, A., Mihatovic, I., Shirakata, Y., Bosshardt, D. D., Schwarz, F., & How to cite this article: Schmitt CM, Brückbauer P, Schlegel Iglhaut, G. (2015). Healing of localized gingival recessions treated KA, Buchbender M, Adler W, Matta RE. Volumetric soft tissue with coronally advanced flap alone or combined with either a resorbable collagen matrix or subepithelial connective tissue graft. A alterations in the early healing phase after peri- implant soft preclinical study. Clinical Oral Investigations, 19, 903–909. https:// tissue contour augmentation with a porcine collagen matrix doi.org/10.1007/s00784-014-1299-x versus the autologous connective tissue graft: A controlled Thoma, D. S., Buranawat, B., Hammerle, C. H., Held, U., & Jung, R. E. clinical trial. J Clin Periodontol. 2021;48:146–163. https://doi. (2014). Efficacy of soft tissue augmentation around dental implants and in partially edentulous areas: a systematic review. Journal org/10.1111/jcpe.13387 of Clinical Periodontology, 41(Suppl 15), S77–S91. https://doi. org/10.1111/jcpe.12220 Thoma, D. S., Hammerle, C. H., Cochran, D. L., Jones, A. A., Gorlach, C., Uebersax, L., Mathes, S., Graf-Hausner, U., & Jung, R. E. DOI: 10.1111/jcpe.13403

CORRIGENDUM

Corrigendum

In Sanz et al. (2020), the authors would like to correct the following errors in the published versión.

• The surname of the sixth author should be corrected from ‘Beglundh’ to ‘Berglundh’ in the author byline and in ORCID section on page 55. Therefore, the correct author list should be:

Mariano Sanz, David Herrera, Moritz Kebschull, Iain Chapple, Søren Jepsen, Tord Berglundh, Anton Sculean, Maurizio S. Tonetti, on behalf of the EFP Workshop Participants and Methodological Consultants.

• The content and format of Table 7 should be identical with Table 3 of ‘Tonetti et al. J Clin Periodontol. 2018;45(Suppl 20): S149–S161’, below is the correct presentation of table:

TABLE 7 Periodontitis stage. Adapted from Tonetti, Greenwell and Kornman (2018)

164 | wileyonlinelibrary.com/journal/jcpe J Clin Periodontol. 2021;48:164–165. CORRIGENDUM | 165

• The content and format of Table 8 should be identical to Table 1B of ‘Papapanou et al. J Clin Periodontol. 2018;45(Suppl 20):S162–S170’, below is the correct presentation of table:

TABLE 8 Periodontitis grade. Adapted from Papapanou et al. (2018)

• On page 19, the subheading under subsection 5.3 should be changed from ‘R1.3’ to ‘R1.4’.

The online version has been corrected.

REFERENCE Sanz, M., Herrera, D., Kebschull, M., Chapple, I., Jepsen, S., Berglundh, T., Sculean, A., Tonetti, M. S., On behalf of the EFP Workshop Participants and Methodological Consultants. (2020). Treatment of stage I-III periodontitis—The EFP S3 level clinical practice guideline. Journal of Clinical Periodontology, 47(Suppl 22), 4–60. https://doi.org/10.1111/jcpe.13290 Tonetti, M. S., Greenwell, H., & Kornman, K. S. (2018). Staging and grading of periodontitis: Framework and proposal of a new classification and case definition. Journal of Clinical Periodontology, 45(Suppl 20), S149–S161. https://doi.org/10.1111/jcpe.12945 Papapanou, P. N., Sanz, M., Buduneli, N., Dietrich, T., Feres, M., Fine, D. H., Flemmig, T. F., Garcia, R., Giannobile, W. V., Graziani, F., Greenwell, H., Herrera, D., Kao, R. T., Kebschull, M., Kinane, D. F., Kirkwood, K. L., Kocher, T., Kornman, K. S., Kumar, P. S., … & Tonetti, M. S. (2018). Periodontitis: Consensus report of workgroup 2 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. Journal of Clinical Periodontology, 45(Suppl 20), S162–S170. https://doi.org/10.1111/jcpe.12946