Evaluation of clinical and radiographic indices as predictors of osteoporotic fractures: a 10-year longitudinal study

Accepted Manuscript Title: Evaluation of clinical and radiographic indices as predictors of osteoporotic fractures: a 10-year longitudinal study Author: Grethe Jonasson, Valter Sundh, Magnus Hakeberg, Margareta Ahlqwist, Lauren Lissner, Dominique Hange PII: DOI: Reference:

S2212-4403(17)31187-2 https://doi.org/10.1016/j.oooo.2017.11.009 OOOO 1886

To appear in:

Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology

Received date: Revised date: Accepted date:

10-6-2017 23-10-2017 3-11-2017

Please cite this article as: Grethe Jonasson, Valter Sundh, Magnus Hakeberg, Margareta Ahlqwist, Lauren Lissner, Dominique Hange, Evaluation of clinical and radiographic indices as predictors of osteoporotic fractures: a 10-year longitudinal study, Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology (2017), https://doi.org/10.1016/j.oooo.2017.11.009. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Evaluation of clinical and radiographic indices as predictors of osteoporotic fractures: a 10-year longitudinal study Grethe Jonasson DDS, PhD, a,b Valter Sundh BSc,c Magnus Hakeberg DDS, PHD, d Margareta Ahlqwist DDS, PhD,e Lauren Lissner MPH, PhD,f Dominique Hange MD, PhD.g. a

Associate professor, Dept. of Behavioral and Community Dentistry, Institute of Odontology

at the Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden. b

Research  Development Unit, Sven Eriksonplatsen 4, SE-503 38 Borås, Sweden.

c

Biostatistician, Dept. of Public Health and Community Medicine, Institute of Medicine,

University of Gothenburg, Gothenburg, Sweden. d

Professor, Dept. of Behavioral and Community Dentistry, Institute of Odontology at the

Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden. e

Associate professor, Dept. of Oral and Maxillofacial Radiology, Institute of Odontology at

the Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden. f

Professor, Section for Epidemiology and Social Medicine, Institute of Medicine, University

of Gothenburg, Gothenburg, Sweden. g

Senior lecturer, Department of Public Health and Community Medicine/Primary Health

Care, Sahlgrenska Academy, University of Gothenburg, Sweden. Running title: FRAX and mandibular cortical bone Number of words in abstract: 198, words in manuscript: 3629, four tables. 2 figures. Corresponding author: Grethe Jonasson, FoU Centrum, Sven Eriksonplatsen 4, SE-50338 Borås, Sweden. TEL +46 709285671. [email protected] Statement of sources of funding for the study. The study was supported by The Health & Medical Care Committee of the Region Västra Götaland, Sweden. Conflict of interest. The authors declare no conflict of interest. 1 Page 1 of 24

Statement of Clinical Relevance. Many patients will suffer osteoporotic fractures that could be prevented if easy low-cost methods for high risk identification were available. Dental radiographs present both trabecular and cortical bone and may be useful for fracture prediction in a general practice setting.

Abstract Objectives: To evaluate two radiographic and three clinical indices as predictors of future osteoporotic fractures. Study design: In a prospective, longitudinal study with a 10-year fracture follow-up, the two radiographic indices, mandibular cortical erosion (normal, mild/moderate erosion, and severe erosion of the inferior cortex) and cortex thickness, were assessed using panoramic radiographs of 411 women, aged 62-78 years. The clinical indices were the fracture assessment tool, FRAX(R), the osteoporosis index of risk, OSIRIS, and the osteoporosis selfassessment tool, OST. Results: The relative risks (RR) for future fracture were significant for FRAX(R)>15%, 4.1 (95% CI: 2.4-7.2), and for severely eroded cortices, 1.7 (95% CI: 1.1-2.8). Cortical thickness <3mm, OSIRIS, and OST were not significant fracture predictors (RR 1.1, 1.4, and 1.5 respectively). For the five tested fracture predictors, Fisher’s exact test gave the following pvalues for differences between fractured and non-fractured groups: FRAX( R) <0.001, cortical erosion 0.023, OST 0.078, OSIRIS 0.206, and cortical thickness 0.678. The area under the curve (AUC) was 0.69 for FRAX(R)>15%, 0.58 for cortical erosion, and 0.52 for cortical thickness. Adding OSIRIS and OST did not change AUC significantly. Conclusion: FRAX(R) and severely eroded cortices predicted fracture but cortical thickness, OSIRIS, and OST did not. 2 Page 2 of 24

Keywords: Bone density; bone fracture; longitudinal; mandible; osteoporosis; population study; prospective; radiography; women.

Introduction Osteoporotic fractures have become a major health problem in the Western world. They occur in forearms, spines, and hips as a result of minor trauma after falls from standing heights, but not in the jaws, where fractures are generally traumatic and not osteoporotic in nature. Measurements of bone mineral density (BMD) by dual X-ray absorptiometry of the hip can establish the diagnosis of osteoporosis, according to the principles recommended by the WHO Scientific Group for Osteoporosis.1 Osteoporosis is strongly associated with fracture risk but BMD is not recommended for screening due to its high costs and low sensitivity.2 Therefore, a WHO group, the Metabolic Bone Disease Group, has developed a fracture assessment tool (FRAX(R)), which uses clinical risk factors and can be used with or without BMD. 3 A FRAX(R) probability of greater than 15% is recommended as the intervention threshold by the Swedish National Board of Health and Welfare, in line with the recommendations of Kanis et al. 3 FRAX(R) is freely accessible on the Internet. All dentists are well trained in interpreting radiographic features and accustomed to working prophylactically. Periapical dental radiographs clearly depict the trabecular bone in both jaws, and panoramic radiographs depict the mandibular inferior cortex. Therefore, many research groups have tried different approaches to predict the risk of osteoporosis, 4-10 and future fracture,11-17 using dental radiographs. Previously we found that a FRAX(R) value greater than 15%, without BMD, was an effective fracture predictor, and mandibular sparse trabeculation had a substantial additive

3 Page 3 of 24

effect. 17 The area under the curve (AUC) in receiver operating characteristic analysis of FRAX(R) >15% was 0.69, as was AUC for sparse trabeculation. 17 Radiographic features of the inferior border of the lower jaw have been used as fracture predictors in cross-sectional studies,11, 16 and in a longitudinal study.14 Furthermore, the relationship between the OSTEODENT Index, cortical thickness combined with clinical information, and hip fracture risk, as determined by FRAX, has been studied.18 The aim of the present investigation was to test the performance of the degree of erosion and the thickness of the mandibular inferior cortex measured on panoramic radiographs, alone and in combination with the three clinical indices, FRAX(R), OSIRIS, and OST, as fracture predictors in a longitudinal study with a 10-year fracture follow-up.

Materials and Methods, The Prospective Population Study of Women in Gothenburg, Sweden is an ongoing longitudinal study initiated in 1968-69 with follow-ups in 1980/81, and 1992/93. The initial sample was representative of women in Gothenburg, chosen randomly from the Revenue Office register, according to date of birth. Ninety percent of the women invited underwent a medical examination, and 97% of these patients also participated in the dental aspect of the study, which corresponded to an 87% participation rate in the dental study in 1968/69.19 The participation rates for the dental studies, calculated in the same way, were 75% in 1980/81, and 64% in 1992/93. The women who refused participation in the first study did not differ significantly from the participants except for long-term survival, which was shown to be lower for the initial refusers.20 The present data includes three age cohorts, born in 1914 (n=28), 1922 (n=171), and 1930 (n=212), i.e. 411 women, between 62 and 78 years of age, examined in 1992/93. The inclusion criteria were a 10-year fracture follow-up and panoramic radiographs from 1992-93 from which it was possible to assess the mandibular cortex. Incident fractures were followed

4 Page 4 of 24

up until 2006, but only 10-year fracture information was used for comparability with the 10year prediction given by FRAX(R). Exclusion criterion was death before 2002/03. The selfreported fractures from questionnaires were merged with fractures from the hospital register from the start and end of the period. Hospital records confirmed nine hip fractures (2.2%), and due to this low number of events, hip fracture incidence was not analyzed separately. In the previous 24-year follow-up,14 using the same three age cohorts from the prospective study of women in Gothenburg, we found that the group with severely eroded cortices increased from 0.5 % in the youngest group (38-year-olds) to 75.4 % in oldest cohort group (78-year-olds). 14 In 1968/69 (38, 46 and 52 years-old participants), 54% of future fractures were found in the group with normal cortices. Twelve years later, when the women were 50, 58, and 66, 62% of the fractures were found in the moderately eroded group. Finally, when the women were 62, 70 and 78, 64% of future fractures in the following 12 years were found in the group with severely eroded inferior cortex.14 Therefore, we chose to use only panoramic radiographs from the latest period (1992/93) when focusing on the accuracy of the mandibular cortex as a fracture predictor. No significant differences were found for cortical thickness between fractured and non-fractured groups. 14 The Ethics Committee of the University of Gothenburg, Sweden approved the study, and participants gave their informed consent Panoramic radiographs. The analogue panoramic radiographs were exposed in 1992/93. The exposure factors had been set individually according to the size of the individual. Small variations had no influence on the visual evaluation of the inferior cortex. Mandibular cortical erosion. Mandibular cortical erosion, also sometimes called mandibular cortical shape, can be categorized by the mandibular cortical index (MCI). The categories in MCI indicate the degree of erosion of the inferior cortex distal to the mental foramen, as

5 Page 5 of 24

illustrated in Figure 1.6, 21, 22 Normal cortex (MCI-1) has an even endosteal margin, moderately eroded cortex (MCI-2) has semilunar defects, and severely eroded cortex (MCI-3) has heavy endosteal cortical porosities. Evaluations of the 411 panoramic radiographs were made consistently by one experienced, blinded dentist (GJ), who classified the right side of each mandible in one of the three categories. Severe cortical erosion (MCI-3) as well as any erosion (MCI-2 +MCI-3) were tested as cut-offs for fracture prediction.5, 6, 8 Mandibular cortical thickness. A research assistant, blinded for fracture outcome and guided by a maxillofacial radiologist, measured inferior cortex thickness on the analogue panoramic radiographs with a specially developed transparent ruler, which took into account the magnification of the panoramic radiograph (Figure 2). Thickness was measured at the mental foramen. Measurements were made consistently on the right side of each mandible of 390 panoramic radiographs. Twenty-one thickness measurements could not be performed due to severe erosion or a very irregular cortex thickness. A thickness of less than 3mm was used as the cut-off for calculations of fracture risk.8,9 FRAX(R). The FRAX(R) probability chart includes age; body mass index (BMI expressed in kg/m2); and clinical risk factors including previous fracture, current smoking, use of oral glucocorticoids, rheumatoid arthritis, and alcohol intake (3 or more units daily). 3, 23 A FRAX(R) value, the 10-year probability of a major osteoporotic fracture (clinical spine, hip, lower arm/wrist, or upper arm fracture) in percent, was calculated using the link: http://www.shef.ac.uk/FRAX/charts/Chart_SWE_ost_wom_bmi.pdf 23 FRAX(R) was calculated for all 411 women. No measurements of BMD were taken on any occasion, and no information was obtained for fractures in parents. A value for BMD can be included in FRAX(R), but the tool can be used without BMD. If a response is missing, the tool calculates the probability of future fracture as if the response was “no.”3

6 Page 6 of 24

Previously, different cut-off values had been examined, and the Swedish recommended cut-off of FRAX(R) >15% was found efficient.17 OSIRIS. The osteoporosis index of risk (OSIRIS) 9 is based on four variables: age, body weight, current hormone replacement therapy (HRT) use, and previous fracture. The index is calculated by adding patient age multiplied by -2, weight in kg multiplied by 2, +2 if a current user of HRT, and -2 with a history of low trauma fracture. An OSIRIS score below -3 indicates a high risk of low BMD; between +1 and -3, an intermediate risk; and greater than +1.0, a low risk. Several cut-off values for calculation of fracture risk were tested. The differences between odds ratios were small and none of them approached significance. The cut-off value for high risk for low BMD (-3) was chosen. 9 OST. The osteoporosis self-assessment tool, OST, 22 is constructed from the integer of 0.2 times the patient’s weight minus the integer of 0.2 times her age. Several cut-off values for calculation of fracture risk were tested. The differences between odds ratios were small and none of them approached significance. The cut-off value was set at < -1.0. 22 Statistical Methods Prediction of fracture. Effect measures for potential risk factors are reported as odds ratios (OR) and relative risks (RR) calculated from binary regression models. The sensitivity, specificity, and area under the receiver operating characteristics (ROC) curve (AUC) were also calculated in order to determine the test’s predictive power. The parameter AUC, calculated from a binary prediction model, is interpreted as the probability that a randomly selected individual from the sample with fracture incidence has a larger predicted probability of fracture than a randomly selected individual from the sample without fracture. Fisher's exact test was used to test differences between fractured and non-fractured groups, reported as p-values, for the five fracture predictors separately.

7 Page 7 of 24

Intraobserver and interobserver agreement. Fifty panoramic radiographs were evaluated twice at four-week intervals by two blinded dentists: one oral and maxillofacial radiologist (MA) and one experienced in classifying bone structure (GJ). The intra- and inter-observer agreement was calculated using weighted kappa statistics. A kappa value of zero represents agreement equivalent to that expected by chance, while a value of 1 represents perfect agreement. In accordance with Landis and Koch,24 the following kappa interpretation scale was used: poor to fair (below 0.4), moderate (0.41–0.60), substantial (0.61–0.80), and almost perfect (0.81–1). The intra- and interobserver agreements for cortical thickness were expressed as the correlation between two measurements of 20 radiographs, performed by a research assistant and an oral and maxillofacial radiologist. Stata version 13.1 (College Station, TX, USA), and Epi Info version 3.5 (Center for Disease Control, Atlanta, GA, USA), were used for statistical calculations.

Results Intraobserver and interobserver agreement. Intraobserver agreement for cortical erosion was substantial to almost perfect (0.67 -0.91), and interobserver agreement was moderate to substantial (0.59 -0.79). Intraobserver agreement for cortical thickness, expressed as the correlation between two measurements, was highly correlated (r=0.88, p<0.001), as was the interobserver agreement (r=0.85, p<0.001). Fracture and FRAX. Sixty-seven women of the total population of 411 (16.3%) sustained a new incident fracture during 1992-2003. Incident fracture rates and FRAX-values for the three age cohorts are presented in Table I. The absolute fracture risk for the 196 women with FRAX(R) >15% was 27% (53 fractures), and for the 215 women with FRAX(R) <15, it was 6.5% (14 fractures; p<0.001). In the smaller

8 Page 8 of 24

group with measured cortical thickness (n=390), the incident fracture rates were 26.0% and 7.0% respectively Mandibular cortical erosion. In the 37 women with cortical erosion MCI-1, fracture risk was 10.8%. In the 158 women with MCI-2, the fracture risk was 12.0%, and in 216 women with MCI-3, 20.4% (p=0.06). In the 374 women with either MCI-2 or MCI-3, the fracture risk was 16.8% and RR= 1.6 (95% CI: 0.6-4.0) The percentages of MCI-3 in the three age groups are presented in Table I. Incident fracture rate and relative risks for fracture in different combinations of risk categories, defined from FRAX >15% and MCI-3 (severe cortical erosion), are presented in Table II together with the predictive power as measured by AUC. AUC values for MCI-3 alone and in combination with OST are also presented in Table II. For those with at least one of the two fracture risk factors (FRAX(R) >15% or MCI-3), 61 of 67 fractures were predicted; sensitivity was 91% and specificity 34%. In the group with both FRAX(R) >15% and MCI-3, 36 of 67 fractures were predicted; sensitivity was 54% and specificity 74%. Cortical thickness. Table I presents mean cortical thickness and incident fracture for the entire group and in the three age cohorts. The 390 individuals with measured cortical thickness sustained 63 fractures (16%) between 1992 and 2003. Mean cortical thickness in the group without fracture was 2.72 + 0.90mm (range 0.90-5.30mm, n=327), and in the fractured group 2.64 + 0.96 (range 0.90-4.50mm, n=63, p=0.54. In the group with cortical thickness < 3mm (n=225), the absolute fracture risk was 16.9%, and in the group with cortical thickness > 3mm (n=165), it was 15.2% (p=0.65). Fracture risk in different risk categories are presented in Table III together with the predictive power as measured by AUC. For those with at least one of two risk factors (FRAX(R) >15% and cortical thickness <3mm), 58 of 63 fractures were predicted, with 92.0% sensitivity and 27.8% specificity. In the group with both risk factors, 29 of 63 fractures were

9 Page 9 of 24

predicted, with a sensitivity of 46% and specificity of 71.8%. AUC values for cortical thickness alone and in combination with FRAX>15% and OSIRIS are also presented in Table III. FRAX>15% alone was the best predictor. When cortical erosion was used along with cortical thickness as fracture predictors in a logistic regression analysis, the odds ratio (OR) for cortical erosion was 2.2 (95% CI: 1.144.21), whereas cortical thickness was not significant (OR=0.81; 95% CI: 0.42-1.54). OSIRIS and OST. Mean values for the entire group as well as for the fractured and nonfractured group are presented in Table IV. Both OSIRIS and OST were significantly correlated to FRAX(R) (r= -0.44, p<0.001, and r =-0.37, p<0.001). The RRs for OSIRIS and OST as fracture predictors were not significant (1.4, 95% CI: 0.85-2.2, and 1.5, 95% CI: 0.982.4). Significance test. For the five tested fracture predictors, Fisher’s exact test gave the following p-values for differences between fractured and non-fractured groups: FRAX (R) p<0.001; cortical erosion p=0.023; OST p=0.078; OSIRIS p=0.206; and cortical thickness p=0.678.

Discussion FRAX(R) >15% indicated a relative risk four times greater than that of a FRAX(R) of less than 15%. This cut-off for FRAX(R) is in line with recommendations for men and women 65-69 years old.25 A FRAX(R) cut-off exceeding 30% has been recommended for 80-year-olds,25 and this seems to be an appropriate limit also in the present sample considering that the mean FRAX(R) value in our study for 78-year-old women was 37.3. Severely eroded cortex (MCI-3) was a significant predictor of future fracture, but a cortical thickness less than 3mm, OSIRIS, and OST were not.

10 Page 10 of 24

Previously, we found that a sparse trabeculation in the mandible and FRAX(R) >15% were strong predictors of future fracture risk. With one risk factor, the relative risk of future fracture was 16-fold over having none, whereas having both was associated with a 23-fold risk of future fracture over having no risk factor.17 In the present study, we found that for those with at least one of the two risk factors, FRAX>15% or MCI-3/cortical thickness<3mm, the relative risks for future fractures were higher (4-fold) than having both (3-fold and 2-fold, respectively). FRAX(R) performed well, despite missing information about parental fracture history and BMD, whereas the other clinical indices, OSIRIS and OST, did not predict fracture. In large studies of 3000 to 6000 men and women, similar AUC values have been found for FRAX(R), OST, OSIRIS, and age as predictors of osteoporosis, 26, 27 but prediction of future fracture differs from osteoporosis prediction because a fall is needed for a fracture. Bone remodeling takes place on endosteal surfaces of cortical and trabecular bone. After menopause, more bone is lost than formed during the remodeling process. Cortical bone loss is slower than that of trabecular bone, which has a larger endosteal surface than does an equal volume of cortical bone. Both cortical and trabecular bone loss can be detected when dental radiographs are carefully studied. When followed on panoramic radiographs (three radiographs in 24 years), cortical bone loss is seen as increased cortical erosion and/or cortical thinning.14 Cortical thickness increases up to the age of 50 and decreased significantly thereafter.14 Increased intertrabecular spaces and less mineralized trabeculae are manifestations of trabecular bone loss.14 Experiments with artificial bone lesions in long bones have indicated that the trabecular pattern apparent in radiographs is formed in the transitional area between the cortical and trabecular bone.28 However, comparisons between the results of image processing of mandibular projection radiographs and micro-computerized tomography of cadaver specimens revealed no evidence that the radiographically observed 11 Page 11 of 24

trabecular bone structure was due mainly to the endosteal surface. 29 Therefore, trabecular bone in the alveolar process, perhaps in contrast to the long bones, probably accounts for most of the visible fine structures in the dental radiographs.29 Cortical bone is not entirely compact but is traversed by Haversian and Volkmann canals with endosteal surfaces for remodeling. Between the ages of 50 and 64, trabecular bone mass may decrease by 22% and cortical bone mass by 4%, but because 80% of skeletal bone is cortical, the absolute bone loss during postmenopausal years may be similar.30 Between ages 65 and 80, four times more bone may be lost than between 50 and 64, and most of it may be cortical, not trabecular. 30 After age 80, bone loss seems to be mainly cortical, because the intracortical surfaces exceed trabecular surfaces. 30 As a result, fractures affecting bone sites with mainly cortical bone may become more common with age than fractures in the vertebrae, which consist mainly of trabecular bone. These calculations of bone loss support the goal of developing cortical structural methods for assessing fracture risk, especially for individuals of advanced age. It is conceivable that evaluation of cortical erosion (MCI) and cortical thickness on panoramic radiographs may perform equally well or better after age 80 than trabeculation on intra-oral radiographs, but this has not as yet been demonstrated. More research is needed for that age group. Mandibular cortical erosion and cortical thickness are mostly evaluated as predictors of osteoporosis, 4-6, 8, 9, 18, 21, 22 but with no BMD measurements in our sample we were unable to test their detection performance. Osteoporotic fractures have been significantly associated with moderately eroded and severely eroded mandibular cortices in two11, 14 of three samples.11, 14, 16 No significant differences were found for cortical thickness between fractured and non-fractured groups in the present investigation or in a previous study.14 In the present study of women, we found that mean cortical thickness was 0.08 mm thinner in the fracture

12 Page 12 of 24

group, whereas cortical thickness was 0.57 mm thinner in women with an osteoporotic fracture than in controls in a case-control study. 11 Sparse trabecular pattern was associated with future fractures both in peri-menopausal and older women.13 Cortical erosion was associated with fracture in older women 14 but cortex thickness was not associated with fracture.14 The better fracture prediction using sparse trabecular bone with or without FRAX(R) 13,14, 17 may be due to the high degree of remodeling in mandibular alveolar processes, probably the highest in the skeleton.31 The alveolar trabecular bone remodels faster than the basal jaw bone and other skeletal bones. 32 Furthermore, the remodeling intensity in the mandible is twice that of the maxilla. 31 Thus, skeletal bone loss may be detected sooner in the alveolar trabecular bone of the mandible than elsewhere. The OSTEODENT Index is a predicted probability of osteoporosis derived from a combination of an automated analysis of the cortical thickness and clinical information (OSIRIS). It produced a high AUC value for osteoporosis prediction,33 but its accuracy for fracture prediction has not been tested as yet. When the cortex is eroded, measurements of cortical thickness are difficult, but great care was applied when measuring cortical thickness manually in this investigation. We did not have the automatic tool developed for that purpose,34 and therefore cannot exclude that the tool performs better than the manual measurements.33 The OSTEODENT Index has been compared with FRAX(R) without BMD.18 The two tools calculated fracture risk similarly for the sample, but no information of sustained fractures was presented. The strengths of this investigation are that it is population-based, has a prospective design, and a long fracture follow-up. The main limitation is that we did not use the automatic tool for cortical thickness, and our results may therefore not be completely fair. Furthermore, a reference object is preferable for measuring cortical thickness. However, all panoramic 13 Page 13 of 24

radiographs had a magnification factor of 1.3, and our Department of Oral and Maxillofacial Radiology has developed a transparent thickness ruler that can be used for different magnifications of the radiographs. Conclusion: In the present sample, FRAX R) and severely eroded cortices predicted fracture but cortical thickness, OSIRIS, and OST did not.

14 Page 14 of 24

References 1. WHO Scientific Group. Prevention and management of osteoporosis. World Health Organ Tech Rep Ser. 2003;921:1-164. 2. Pasco JA, Seeman E, Henry MJ, Merriman EN, Nicholson GC, Kotowicz MA. The population burden of fractures originates in women with osteopenia, not osteoporosis. Osteoporos Int. 2006;17:1404–1409. DOI: 10.1007/s00198-0060135-9 3. Kanis JA, Johnell O, Oden A, Johansson H, McCloskey E. FRAX and the assessment of fracture probability in men and women from the UK. Osteoporos Int. 2008;19:385-397. DOI: 10.1007/s00198-007-0543-5 4. Horner K, Devlin H, Harvey L. Detecting patients with low skeletal bone mass. J Dent. 2002;30:171-175. https://doi.org/10.1016/S0300-5712(02)00010-6 5. White SC, Taguchi A, Kao D, Wu S, Service SK, Yoon D, Suei Y, Nakamoto T, Tanimoto K. Clinical and panoramic predictors of femur bone mineral density. Osteoporos Int. 2005;16:339–346. DOI: 10.1007/s00198-004-1692-4 6. Taguchi A, Tsuda M, Ohtsuka M, Kodama I, Sanada M, Nakamoto T, Inagaki K, Noguchi T, Kudo Y, Suei Y, Tanimoto K, Bollen AM. Use of dental panoramic radiographs in identifying younger postmenopausal women with osteoporosis. Osteoporos Int. 2006;17:387–394. DOI: 10.1007/s00198-005-2029-7 7. Jonasson G, Jonasson L, Kiliaridis S. Changes in the radiographic characteristics of the mandibular alveolar process in dentate women with varying bone mineral density: a 5-year prospective study. Bone 2006;38:714-721. DOI: 10.1016/j.bone.2005.10.008 8. Devlin H, Karayianni K, Mitsea A, Jacobs R, Lindh C, van der Stelt P, Marjanovic E, Adams J, Pavitt S, Horner K. Diagnosing osteoporosis by using dental 15 Page 15 of 24

panoramic radiographs: the OSTEODENT project. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2007;104:821-828. DOI: 10.1016/j.tripleo.2006.12.027 9. Karayianni K, Horner K, Mitsea A, Berkas L, Mastoris M, Jacobs R, Lindh C, van der Stelt PF, Harrison E, Adams JE, Pavitt S, Devlin H. Accuracy in osteoporosis diagnosis of a combination of mandibular cortical width measurement on dental panoramic radiographs and a clinical risk index (OSIRIS): the OSTEODENT project. Bone 2007;40:223-229. DOI: 10.1016/j.bone.2006.07.025 10. Lindh C, Horner K, Jonasson G, Olsson P, Rohlin M, Jacobs R, Karayianni K, van der Stelt P, Adams J, Marjanovic E, Pavitt S, Devlin H. The use of visual assessment of dental radiographs for identifying women at risk of having osteoporosis: the OSTEODENT project. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2008;106:285-293. DOI: 10.1016/j.tripleo.2007.09.008 11. Bollen AM, Taguchi A, Hujoel PP, Hollender LG. Case-control study on selfreported osteoporotic fractures and mandibular cortical bone. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2000;90:518-524. DOI: 10.1067/moe.2000.107802 12. White SC, Atchison KA, Gornbein JA, Nattiv A, Paganini-Hill A, Service SK, Yoon DC. Change in mandibular trabecular pattern and hip fracture rate in elderly women. Dentomaxillofac Radiol. 2005;34:168–174. DOI: 10.1259/dmfr/32120028 13. Jonasson G, Sundh V, Ahlqwist M, Hakeberg M, Björkelund C, Lissner L. A prospective study of mandibular trabecular bone to predict fracture risk: a low-cost screening tool in the dental clinic. Bone 2011;49:873-879. DOI: 10.1016/j.bone.2011.06.036

16 Page 16 of 24

14. Jonasson G, Sundh V, Hakeberg M, Hassani-Nejad A, Lissner L, Ahlqwist M. Mandibular bone changes in 24 years and skeletal fracture prediction. Clin Oral Investig. 2013;17:565-572. DOI: 10.1007/s00784-012-0745-x 15. Hassani-Nejad A, Ahlqwist M, Hakeberg M, Jonasson G. Mandibular trabecular bone as fracture indicator in 80-year-old men and women. Eur J Oral Sci. 2013;121:525-531. DOI: 10.1111/eos.12087 16. Yamada S, Uchida K, Iwamoto Y, Sugino N, Yoshinari N, Kagami H, Taguchi A. Panoramic radiography measurements, osteoporosis diagnoses and fractures in Japanese men and women. Oral Dis. 2015;21:335-341. DOI: 10.1111/odi.12282 17. Sundh V, Hange D, Ahlqwist M, Hakeberg M, Lissner L, Jonasson G. FRAX and mandibular sparse trabeculation as fracture predictors: a longitudinal study 19922002. Eur J Oral Sci. 2017;125:135–140. DOI: 10.1111/eos.12341 18. Horner K, Allen P, Graham J, Jacobs R, Boonen S, Pavitt S, Nackaerts O, Marjanovic E, Adams JE, Karayianni K, Lindh C, van der Stelt P, Devlin H. The relationship between the OSTEODENT index and hip fracture risk assessment using FRAX. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2010;110:243249. DOI: 10.1016/j.tripleo.2010.03.035 19. Ahlqwist M, Bengtsson C, Gröndahl HG, Lapidus L. Social factors and tooth loss in a 12-year follow-up study of women in Gothenburg, Sweden. Community Dent Oral Epidemiol .1991;19:141–146. DOI: 10.1111/j.1600-0528.1991.tb00129.x 20. Bengtsson C, Ahlqwist M, Andersson K, Björkelund C, Lissner L, Söderström M. The Prospective Population Study of Women in Gothenburg, Sweden, 1968–69 to 1992–93. A 24-year follow-up study with special reference to participation, representativeness, and mortality. Scand J Prim Health Care 1997;15:214–219.

17 Page 17 of 24

21. Klemetti E, Kolmakov S, Kröger H. Pantomography in assessment of the osteoporosis risk group. Scand J Dent Res. 1994;102:68–72. 22. Taguchi A, Suei Y, Sanada M, Ohtsuka M, Nakamoto T, Sumida H, Ohama K, Tanimoto K. Validation of dental panoramic radiography measures for identifying postmenopausal women with spinal osteoporosis. AJR Am J Roentgenol. 2004;183:1755-1760. DOI: 10.2214/ajr.183.6.01831755 23. http://www.shef.ac.uk/FRAX/charts/Chart_SWE_ost_wom_bmi.pdf 24. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. 25. Kanis JA, McCloskey EV, Johansson H, Strom O, Borgstrom F, Oden A; National Osteoporosis Guideline Group. Case finding for the management of osteoporosis with FRAX assessment and intervention thresholds for the UK. Osteoporos Int. 2008;19:1395-1408. DOI: 10.1007/s00198-008-0712-1 26. Rubin KH, Abrahamsen B, Friis-Holmberg T, Hjelmborg JV, Bech M, Hermann AP, Barkmann R, Glüer CC, Brixen K. Comparison of different screening tools (FRAX®, OST, ORAI, OSIRIS, SCORE and age alone) to identify women with increased risk of fracture. A population-based prospective study. Bone 2013;56:1622. DOI: 10.1016/j.bone.2013.05.002 27. Ensrud KE, Lui LY, Taylor BC, Schousboe JT, Donaldson MG, Fink HA, Cauley JA, Hillier TA, Browner WS, Cummings SR. Study of Osteoporotic Fractures Research Group. A comparison of prediction models for fractures in older women: is more better? Arch Intern Med. 2009;169:2087-94. DOI: 10.1001/archinternmed.2009.404 28. Van der Stelt PF. Experimentally produced bone lesions. Oral Surg Oral Med

18 Page 18 of 24

Oral Pathol 1985; 59: 306-12. doi.org/10.1016/0030-4220(85)90172-0 29. Couture RA, Whiting BR, Hildebolt CF, Dixon DA. Visibility of trabecular structures in oral radiographs. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2003; 96: 764-71. DOI: 10.1016/S1079210403005146 30. Zebaze RM, Ghasem-Zadeh A, Bohte A, Iuliano-Burns S, Mirams M, Price RI, Mackie EJ, Seeman E. Intracortical remodelling and porosity in the distal radius and post-mortem femurs of women: a cross-sectional study. Lancet 2010;375:1729–1236. DOI: 10.1016/S0140-6736(10)60320-0 31. Huja SS, Fernandez SA, Hill KJ, Li Y. Remodeling dynamics in the alveolar process in skeletally mature dogs. Anat Rec Discov Mol Cell Evol Biol. 2006;288:1243–1249. DOI: 10.1002/ar.a.20396 32. Marx RE, Cillo JE Jr, Ulloa JJ. Oral Bisphosphonate-Induced Osteonecrosis: Risk Factors, Prediction of Risk Using Serum CTX Testing, Prevention, and Treatment. J Oral Maxillofac Surg. 2007;65:2397-2410. DOI: 10.1016/j.joms.2007.08.003 33. Devlin H, Allen P, Graham J, Jacobs R, Nicopoulou-Karayianni K, Lindh C, Marjanovic E, Adams J, Pavitt S, van der Stelt P, Horner K. The role of the dental surgeon in detecting osteoporosis: the OSTEODENT study. Br Dent J. 2008;204:E1–E6. DOI: 10.1038/sj.bdj.2008.317 34. Devlin H, Allen PD, Graham J, Jacobs R, Karayianni K, Lindh C, van der Stelt PF, Harrison E, Adams JE, Pavitt S, Horner K. Automated osteoporosis risk assessment by dentists: a new pathway to diagnosis. Bone 2007;40:835–842. DOI: 10.1016/j.bone.2006.10.024

19 Page 19 of 24

Figure 1. Reference panoramic radiographs for mandibular cortical index (MCI) A. A normal cortex (MCI-1, left) with an even endosteal margin. B. A moderately eroded cortex (MCI-2, middle) with semilunar defects. C. A severely eroded cortex (MCI-3, right) with heavy endosteal cortical porosities. Figure 2. Measurement of cortical thickness on panoramic radiographs with a transparent ruler that takes into account the magnification factor (1.3).

20 Page 20 of 24

Table I. Fracture percentage, mean FRAX value, severely eroded cortex (MCI-3), and mean cortical thickness (CT) listed as the total for all groups and for different age groups. The pvalues are calculated with Fisher’s exact test. Total

62-yr

70-yr

78-yr

(n=411)

(n=212)

(n=171)

(n=28)

p

16.3

14.1

17.6

25.0

0.291

18.3 (6.5-56)

12.1 (6.5-20)

22.9 (8.3-51)

37.3 (25.-56)

<0.001*

52.6

40.1

64.9

71.4

<0.001*

2.7 (0.9-5.3)

2.8 (0.9-5.0)

2.6 (0.9-5.3)

2.5 (1.1-2.5)

0.299

Fracture % FRAX % (range) MCI-3 % CT mm (range)

*= significant difference between the four age groups

Table II. Absolute and relative fracture risk in women in combinations of risk categories, defined from FRAX >15%, severe cortical erosion (MCI-3), and OST (411 assessed and 67 fractures).

(n) Fracture %

Relative risk

AUC (95% CI)

FRAX>15 vs. FRAX < 15%

(196) 27%

4.1 (2.4-7.2)

0.69 (0.63-0.74)

MCI-3 vs. not MCI-3

(216) 20%

1.7 (1.08-2.8)

0.58 (0.51-0.64)

Either FRAX > 15% or MCI-3

(288) 21%

4.3 (1.9-9.8)

0.63 (0.58-0.67)

Both FRAX > 15% and MCI-3

(124) 29%

2.7 (1.7-4.1)

0.64 (0.57-0.70)

Both OST and MCI-3

(82) 24%

1.7 (1.1-2.7)

0.56 (0.50-0.62)

Fracture predictors

21 Page 21 of 24

Table III. Absolute and relative fracture risk in women in combinations of risk categories, defined from FRAX >15% and cortex thickness (CT) less than 3mm (390 assessed and 63 fractures).

(n) Fracture %

Relative Risk

AUC (95%)

FRAX>15% vs. FRAX<15%

(189) 26%

3.7 (2.1-6.5)

0.67 (0.62-0.73)

CT< 3mm vs. CT>3mm

(225) 17%

1.1 (0.7-1.8)

0.52 (0.45-0.58)

Either FRAX> 15% or CT<3mm

(293) 20%

3.8 (1.6-9.3)

0.60 (0.56-0.64)

Both CT<3 and FRAX>15%

(121) 24%

1.9 (1.2-3.0)

0.59 (0.52-0.66)

Both CT<3 and OSIRIS

(118) 20%

1.3 (0.8-2.0)

0.53 (0.47-0.60)

Fracture predictors

Table IV. Mean values with standard deviation and range of the three clinical indices (FRAX, OSIRIS and OST) for the total group, for the group with incident fracture, and for the group without fracture.

FRAX

Total group

Fractured group

Not fractured

18.3+9.7 (6.5-56)

22.9+9.8 (12.0-56)

17.4+9.5 (6.5-56)***

OSIRIS

0.25+22.6 (-61.6 to 114)

-2.16+20.9 (-61.6 to 43)

0.73+23.0 (-53.8 to 114)ns

OST

-0.11+2.26 (-6.4 to 11.4)

-0.46+2.1 (-6.4 to 4.1)

-0.04+2.3 (-5.4 to 11.4) ns

***: p<0.001, ns: p>0.05.

22 Page 22 of 24

Fig1_bestsetConverted.png

23 Page 23 of 24

Fig2_bestsetConverted.png

24 Page 24 of 24