Osteoporotic Vertebral Fracture Prevalence Varies Widely Between Qualitative and Quantitative Radiological Assessment Methods: The Rotterdam Study

Abstract

Accurate diagnosis of vertebral osteoporotic fractures is crucial for the identification of individuals at high risk of future fractures. Different methods for radiological assessment of vertebral fractures exist, but a gold standard is lacking. The aim of our study was to estimate statistical measures of agreement and prevalence of osteoporotic vertebral fractures in the population-based Rotterdam Study, across two assessment methods. The quantitative morphometry assisted by SpineAnalyzer® (QM SA) method evaluates vertebral height loss that affects vertebral shape whereas the algorithm-based qualitative (ABQ) method judges endplate integrity and includes guidelines for the differentiation of vertebral fracture and nonfracture deformities. Cross-sectional radiographs were assessed for 7582 participants aged 45 to 95 years. With QM SA, the prevalence was 14.2% (95% CI, 13.4% to 15.0%), compared to 4.0% (95% CI, 3.6% to 4.5%) with ABQ. Inter-method agreement according to kappa (k) was 0.24. The highest agreement between methods was among females (k ¼ 0.31), participants age >80 years (k ¼ 0.40), and at the L 1 level (k ¼ 0.40). With ABQ, most fractures were found at the thoracolumbar junction (T 12 -L 1 ) followed by the T 7 -T 8 level, whereas with QM SA, most deformities were in the mid thoracic (T 7 -T 8 ) and lower thoracic spine (T 11 -T 12 ), with similar number of fractures in both peaks. Excluding mild QM SA deformities (grade 1 with QM) from the analysis increased, the agreement between the methods from k ¼ 0.24 to 0.40, whereas reexamining mild deformities based on endplate depression increased agreement from k ¼ 0.24 to 0.50 (p <0.001). Vertebral fracture prevalence differs significantly between QM SA and ABQ; reexamining QM mild deformities based on endplate depression would increase the agreement between methods. More widespread and consistent application of an optimal method may improve clinical care.

The Radiology of Osteoporotic Vertebral Fractures Revisited

“…Nevertheless, the main source of disagreement between methods as compared by Jiang and colleagues was the large number of prevalent fractures identified by the purely morphometric application of the GSQ method but not by either the relatively loose qualitative morphological criteria or the more rigorous formalism of the ABQ method of diagnosis. They found “many of these were mild thoracic wedge fractures identified as non‐fracture deformities or normal variants by the other two methods.” They go on to note that “bone density was lower in the women with fractures as designated consistently by all three methods [ie, qualitative, semiquantitative and ABQ], than in women with discrepant [G]SQ fractures.” Differences in the way various methods of fracture definition are applied will inevitably lead to a certain amount of disagreement between methods, but the disproportionately large number of discrepant [G]SQ fractures identified in the Jiang and colleagues study raises questions about the true nature of these “deformities.” Notably, the prevalence of morphologic vertebral fractures was very similar in the three recent studies described here . A comparison of the methods is summarized in Table , and some examples of radiographs illustrating the different diagnostic scenarios across methods are presented in Fig.…”

“…(43–45) Grade 1 morphometric deformities (as defined by the GSQ method) were only weakly associated with incident OVFs and not at all with incident nonvertebral fractures as previously observed by Johansson and colleagues . In the Rotterdam Study, fractures scored by quantitative morphometry, when including grade 1 in the definition, were weakly associated with incident nonvertebral fractures . Contrastingly, prevalent grade 1 OVFs defined by a modified morphologic ABQ (mABQ) method were associated with a significant occurrence of both incident OVFs and incident nonvertebral fragility fractures (Table ).…”

“…However, as Jiang and colleagues observed, morphologic fractures may occur in vertebrae that would be graded morphometrically as normal (grade 0), so that any grading system for future use has to provide for that contingency (scored as ABQ 0). In the Rotterdam Study, 35% of the ABQ fractures were observed in vertebrae without morphometric compromise . We believe that fracture grading should be preserved, perhaps using the Genant tool but only when allied to recognition of the qualitative signs of fracturing.…”

“…Until very recently, there remained no satisfactory evidentiary basis to justify the preferential use of morphometric, morphologic, or mixed criteria in any one or other guise, for application in clinical practice . However, Oei and colleagues, Lentle and colleagues, and Deng and colleagues have all recently reported comparisons of the morphological (qualitative) and quasi‐morphometric (quantitative) approaches to the diagnosis of OVF in large prospective studies in participants of both sexes. These studies’ population sizes were 7582, 6236, and 3907 subjects, respectively.…”

“…No such grading is implicit in the qualitative, morphological models described here, although in the CaM os study the mABQ tool was blended with the GSQ grading system . In the Rotterdam Study fractures scored by ABQ and quantitative morphometry grade 2 and 3 held stronger associations with fractures (both vertebral and nonvertebral), as well as with BMD, age, and sex. These data suggest the clinical utility of a grading system.…”