Panoramic measures for oral bone mass in detecting osteoporosis: a systematic review and meta-analysis

Abstract

Different quantitative and qualitative indices calculated on oral panoramic radiographs have been proposed as useful tools to screen for reduced skeletal bone mineral density (BMD). Our aim was to systematically review the literature on linear and qualitative panoramic measures and to assess the accuracy of these indices by performing a meta-analysis of their sensitivity and specificity. The Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement was followed. Fifty studies were included in the qualitative appraisal and 19 were considered for meta-analysis. The methodological quality of the retrieved studies, assessed with the QUADAS-2 tool, was on average low. Three indices were reported by most of the studies: mandibular cortical width, panoramic mandibular index, and the Klemetti index. Mandibular cortical width presented with a better accuracy in excluding osteopenia/osteoporosis (specificity), since patients with a cortical width more than 4 mm had a normal BMD in 90% of the cases. Almost all studies used a cutoff of 0.3 for the panoramic mandibular index, resulting in an estimated sensitivity and specificity in detecting reduced BMD, respectively, of 0.723 (SE 0.160; 95% confidence interval [CI], 0.352-0.926) and 0.733 (SE 0.066; 95% CI, 0.587-0.841). The presence of any kind of mandibular cortical erosion gave an estimated sensitivity and specificity in detecting reduced BMD, respectively, of 0.789 (SE 0.031; 95% CI, 0.721-0.843) and 0.562 (SE 0.047; 95% CI, 0.47-0.651) and a sensitivity and specificity in detecting osteoporosis, respectively, of 0.806 (SE 0.105; 95% CI, 0.528-0.9200) and 0.643 (SE 0.109; 95% CI, 0.417-0.820). The mandibular cortical width, panoramic mandibular index, and Klemetti index are overall useful tools that potentially could be used by dentists to screen for low BMD. Their limitations are mainly related to the experience/agreement between different operators and the different image quality and magnification of the panoramic radiographs.

Keywords: bone density; bone resorption; dental radiography; metabolic bone disease; panoramic radiography; pathologic bone demineralization.

Conflict of interest statement

The authors received no financial support and declare no potential conflicts of interest with respect to the authorship and/or publication of this article.

© International & American Associations for Dental Research.

Figures

Figure 1.

Four-phase flow diagram of the article selection procedure, according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement (Moher et al. 2009).

Figure 2.

Linear panoramic indices: mandibular cortical width (or mental index or mandibular cortical thickness) = b (Horner and Devlin 1998); panoramic mandibular index (PMI) = b/a (Benson et al. 1991); gonial index (GI) = d (Bras et al. 1982); antegonial index (AI) = e (Ledgerton et al. 1999); mandibular ratio (M/M) = c/a (Ortman et al. 1989).

Figure 3.

The bar chart shows that the risk of bias comes mainly from the domains of index test and patient selection, with flow and timing having the highest values for unclear risk.

Figure 4.

The bar chart shows that most of the concerns regarding applicability come from index test and patient selection domains.

Figure 5.

Forest plot of different indices at different cutoffs (RevMan output). The Marandi et al. (2010) study is reported twice for mandibular cortical width (MCW) (cutoff ≤3 mm and ≥4 mm) and for the panoramic mandibular index (PMI), since they reported 2 different 2 × 2 tables within the same cutoff range. However, when estimating the overall sensitivity and specificity and receiver operating characteristic (ROC) curve plots, each 2 × 2 table was used in a different analysis (see Fig. 6). BMD, bone mineral density; CI, confidence interval; FN, false negative; FP, false positive; TN, true negative; TP, true positive.

Figure 6.

Hierarchical summary receiver operating characteristic (HSROC) curves of the studies plotted in Figure 5 (STATA output). (a) Receiver operating characteristic (ROC) curve plot for 3 ≤ mandibular cortical width (MCW) ≥ 4 mm in detecting reduced bone mineral density (BMD): estimated sensitivity, 0.42 (SE 0.168; 95% confidence interval [CI], 0.158–0.737); estimated specificity, 0.93 (SE 0.042; 95% CI, 0.79–0.979). These estimates were obtained when considering the cutoff of ≤3.69 mm for Marandi et al (2010). (a2) Same as a, but for Marandi et al. (2010) we considered the cutoff of ≤4 mm. Estimated sensitivity, 0.43 (SE 0.164; 95% CI, 0.168–0.737); estimated specificity, 0.905 (SE 0.041; 95% CI, 0.789–0.960). (b) ROC plot for 4 < MCW > 5 mm in detecting reduced BMD; estimated sensitivity, 0.602 (SE 0.101; 95% CI, 0.398–0.775); estimated specificity, 0.708 (SE 0.064; 95% CI, 0.568–0.817). (c) ROC plot for panoramic mandibular index (PMI) ≤0.3 in detecting reduced BMD: estimated sensitivity, 0.723 (SE 0.160; 95% CI, 0.352–0.926); estimated specificity, 0.733 (SE 0.066; 95% CI, 0.587–0.841). (c2) Same as c, but for Marandi et al. (2010), we considered the cutoff of ≤0.33 instead of ≤0.32. Estimated sensitivity, 0.729 (SE 0.167; 95% CI, 0.338–0.934); estimated specificity, 0.707 (SE 0.067; 95% CI, 0.561–0.820). (d) ROC plot for eroded cortex (C2+C3) in detecting reduced BMD: estimated sensitivity, 0.789 (SE 0.031; 95% CI, 0.721–0.843); estimated specificity, 0.562 (SE 0.047; 95% CI, 0.47–0.651). (e) ROC plot for eroded cortex (C2+C3) in detecting osteoporosis: estimated sensitivity, 0.806 (SE 0.105; 95% CI, 0.528–0.9200); estimated specificity, 0.643 (SE 0.109; 95% CI, 0.417–0.820).