- Email: [email protected]
Vertebral Morphometry
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. -, no. -, 1e6, 2015 Ó Copyright 2015 Published by Elsevier Inc. on behalf of The International Society for Clinical Densitometry 1094-6950/-:1e6/$36.00 http://dx.doi.org/10.1016/j.jocd.2015.08.005
Original Article
Vertebral Morphometry Sharon H. Chou,1 and Tamara Vokes*,2 1
Division of Endocrinology, Diabetes, and Hypertension, Brigham and Women’s Hospital, Boston, MA, USA; and 2Section of Adult & Pediatric Endocrinology, Diabetes, Metabolism, The University of Chicago, Chicago, IL, USA
Abstract There is as yet no agreement about the criteria by which to arrive at an imaging diagnosis of a vertebral fracture. Because high-grade fractures result in alterations in vertebral shape, 1 possible avenue of diagnosis has been to quantitate changes in vertebral shape. The result has been a variety of methods for the relative and absolute measurements of vertebral dimensions. Such measurements have also lent themselves to automated computed analysis. The number of techniques reflects the absence of any consensus about the best. The semiquantitative technique proposed by Genant has become the most widely used and has served the field well for comparative purposes. Its use in higher grade fractures has been widely endorsed, if some concepts (e.g., short vertebral heightevertebrae) are at variance with lower grades of fracturing. Vertebral morphometry may be the only recourse in high volume epidemiological and interventional studies. Key Words: Vertebral fractures; quantitative morphometry; semiquantitative morphometry.
Quantitative morphometry (QM) methods involve actual measurement of vertebral dimensions with the most commonly used method being 6-point morphometry. In addition to the latter, there are other methods based on the shape of the vertebral body (2,3). More recently, QM by parametric modeling of vertebral bodies in 3 dimensions has been developed (4). In this article, we will focus on 6-point morphometry as most of reported studies have used this method.
Introduction The diagnosis of vertebral fractures is often based on changes in shape due to a compressive force along the superioreinferior axis resulting in a fracture of the vertebral end plate. However, due to the nature of the changes, there is no consensus about the definition or grading of fractures. Broadly, the methods for diagnosing vertebral fractures can be classified as qualitative, quantitative, and semiquantitative (SQ). Qualitatively, a radiologist or other trained observer inspects the spine image(s) and decides whether a given vertebra is fractured. The diagnosis of vertebral fracture by this method is based on several factors such as changes in shape including deformity of the end plate, buckling of cortices, and loss of vertebral height. SQ methods, of which the method of Genant et al (1) has been the most widely used, rely on visual identification of abnormal vertebral shape and comparison of its appearance to a chart (Fig. 1).
Technique: Six-Point Morphometry On lateral spine images (radiographs or dual-energy X-ray absorptiometry spine images, termed vertebral fracture assessment [VFA]), points are manually or digitally placed on each of the 4 corners of the vertebral body, and 2 additional points are placed in the center of the upper and lower end plates (Fig. 2A). These 6 points are used to determine the anterior, mid, and posterior vertebral heights of each vertebra (Fig. 2A). Using these points, each vertebra is then assigned a type and grade (severity) of fracture. The type of deformity is based on lower than expected ratio of particular vertebral heights: wedge for decreased ratio of anterior-toposterior heights (Fig. 3A), biconcave for decreased ratio of mid-to-posterior heights, and compression (or crush) for decreased ratio of posterior height to the posterior height of
Received 10/24/14; Accepted 08/12/15. *Address correspondence to: Tamara Vokes, MD, Section of Adult & Pediatric Endocrinology, Diabetes, Metabolism, The University of Chicago, 5841 South Maryland Avenue, MC 1027, Chicago, IL 60637. E-mail: [email protected]
1
2
Chou and Vokes
Fig. 1. Genant semiquantitative scale.
adjacent vertebrae (Fig. 3B). However, there is normal variation in the shape of vertebral bodies: mid-thoracic vertebrae may be slightly wedge shaped, and normal lumbar vertebrae may have a biconcave shape (5). How to define what degree of deformity constitutes a fracture and how to assign it a grade are a matter of debate with several methods studied (see in the following section) but no consensus reached. Although QM is intended to be quantitative, point placement is subjective and not always straightforward. Osteophytes, Schmorl nodes, and the uncinate process at the posterosuperior border of the thoracic vertebrae should be excluded for point placement (6; Fig. 4A). Point placement may be challenging for images with these non-fracture deformities and for those with projectional variations resulting from radiographic technique (Figs. 2B and 4B). Despite subjectivity and difficulty with point placement, the accuracy and precision of QM are reasonably good for well-trained observers. Accuracy was assessed in a small ex vivo study using 9 cadaveric vertebral columns (7). Anterior, mid, and posterior vertebral heights were assessed by direct measurement and by QM performed on radiographs, as well as on dual-energy X-ray absorptiometry images. Both QM approaches were strongly correlated (r2 5 0.99) with direct measurements. Reproducibility of QM was also found to be good in 400 postmenopausal women with low
Fig. 2. Placement of morphometric points in a vertebra with orthogonal projection (A) and one with obliquity (B). Vertebral heights: anterior (a), mid-vertebral (m), and posterior (p). From http://www.advances-in-medicine.net/2011/ 12/conventional-x-rays-in-the-diagnosis-of-osteoporosis-morp hometric-vertebral-fracture-analysis.html.
Fig. 3. Examples of spine radiographs with fractures as indicated by the white arrows: wedge fracture at T7danterior height is lower than posterior height (A); crush fracture at T5dposterior height of T5 is lower than the posterior heights of T4 and T6 (B). Crush fractures typically have decreased vertebral heights throughout the vertebra. Please note that point placement can be debated in these examples, exemplifying some of the uncertainties in vertebral morphometry. bone mass and at least 1 vertebral fracture with coefficients of variation of 2%e4% for both intraobserver and interobserver comparisons of vertebral heights (8).
Defining Prevalent Vertebral Fractures There is no consensus about defining vertebral fractures by QM. Vertebral fractures can be defined as a reduction in vertebral height compared with other heights within the vertebra or to a normal height as determined by a reference population. Furthermore, height differences that constitute a fracture can be based on percent reduction or standard deviation (SD) from the norm. The Melton Method defined a vertebral fracture as any ratio of anterior-to-posterior height, mid-to-posterior height, or posterior-to-posterior height of adjacent vertebrae that is less than 0.85 (9). To account for natural variations in vertebral shape and size at each level, the individual’s ratio may be
Fig. 4. Challenges to point placement on spine radiographs: avoidance of labeling osteophytes (A) and difficulty (subjectivity) of point placement with obliquity (B). Please note that point placement in these examples are debatable.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Vertebral morphometry adjusted by a vertebra-specific factor that normalizes it to a reference population without vertebral fractures. As expected, when the ratios were adjusted, the prevalence of vertebral fractures in 200 women aged 50 yr decreased from 83% to 26%, closer to the prevalence of 28% as determined by qualitative reading. The Eastell Method defined fracture simply as a ratio of vertebral heights that is more than 3 SD below the normal mean for that vertebral level (10). The normal mean was derived from 52 healthy women without vertebral fractures. Fractures of 3e4 SD are classified as grade 1, and fractures greater than 4 SDs are grade 2. The McCloskey Method aimed to limit the number of false-positive results with the Eastell Method by also requiring that the ratio of any measured vertebral height to predicted posterior height based on adjacent normal vertebrae must also be greater than 3 SD below the mean (11). The Black Method modified the method of Eastell by using a reference population of 2992 women aged 65e70 yr from the Study of Osteoporotic Fractures (SOF) that included women with vertebral fractures but excluded the outermost data by statistical trimming (12). Fractures were defined as values greater than 3 SD below the mean. This method assumes that the ratios among non-fractured vertebral bodies obeys a Gaussian or normal distribution, which may not be the case (13), and the prevalence of abnormal vertebrae at any level is !10%. Other similar methods have been described by Jiang et al (14), Minne et al (15), and others. Most of these quantitative methods require comparison to a reference population. This population should be similar to the study population in terms of age and demographics (16e18). Separate reference values for men and women should also be used (16). The reference range can be derived from a similar population with or without vertebral deformities. If the reference population contains some women (!30%) with vertebral deformities, the statistical methods described by Black et al (12) should be considered. Statistical trimming of a reference population with high prevalence of vertebral fractures, however, will increase false negatives (14). Finally, the source of the image (conventional spinal radiographs vs VFA images) should also be consistent between the study and reference populations (19).
The Genant Semiquantitative Method Because QM is strictly based on vertebral measurements, it may misclassify as fractures some non-fracture deformities such as those resulting from Scheuermann disease or osteoarthritis (20). To avoid such errors and simplify the diagnosis of vertebral fractures, Genant et al (1) introduced an SQ method. With this approach, each vertebra from T4 to L4 is assessed visually for height loss. Signs of fracture are described and include lack of end plate parallelism, end plate depression, buckling of cortical margin, and loss of vertical continuity with adjacent vertebrae, but these are not required for diagnosis. Once a vertebra is designated as fractured, then
3 the grade is assigned by visually comparing it to a chart (Fig. 1): height loss of approximately 20%e25% is labeled as grade 1 (mild), 25%e40% as grade 2 (moderate), and O40% as grade 3 (severe) compared with other heights of the same and/or neighboring vertebrae. Unlike quantitative methods, this approach does not require a reference population for comparison. However, this approach does require training to distinguish fracture from non-fracture deformity or normal variant and suffers from overlap with the short vertebral height variant (21). Despite this, intraobserver variability has been found to be low with percentage agreement between repeat readings of 93% for inexperienced and 97% for experienced observers (1). Interobserver agreement is similarly high at 94% (1,8).
Algorithm-Based Qualitative Assessment The algorithm-based qualitative (ABQ) method is a more recent development resulting from the possible shortcomings of QM and SQ method of Genant (21). Instead of focusing on reductions in vertebral height, the ABQ method focuses on the presence of end plate deformity. The diagnosis of an osteoporotic vertebral fracture by the ABQ method requires an end plate depression, which is symmetrical, concave, involving the whole end plate depression within the rim, and is not associated with trauma, tumor, or metabolic disease. Vertebrae with end plate depressions not meeting additional criteria or vertebrae with no end plate fracture but short vertebral height are classified as non-fracture deformities, developmental variant, non-osteoporotic fracture, or other condition. Prevalence of vertebral fractures in postmenopausal women is lower with ABQ than SQ methods (7% vs 24%) (21) but was not much different in older men (10% vs 13%; 22).
Mild Vs Severe Fractures Mild fractures are controversial in terms of diagnosis and clinical implications. Mild fractures are deformities resulting in a height reduction to 3e4 SD below the mean (about 20% height loss), and more severe fractures are 4 SD below the mean. Depending on the cutoff criteria (and reference population), the prevalence of vertebral fractures can range from 7% to 90% (23). In general, when including mild fractures (3 SD and 15% reduction), the prevalence of vertebral fractures in postmenopausal women is about 20%e25% (9,10,12,24,25). More stringent criteria (4 SD) decrease the prevalence to about 10% (10,25). Unlike 4 SD fractures, 3 SD vertebral fractures are not associated with increased back pain, disability, height loss, or lower lumbar spine bone mineral density (BMD) (20,26e28). Furthermore, 3 SD fractures as determined by methods by Eastell and Black and fractures of 15%e20% reduction in height are not predictive of future vertebral fractures (23). These deformities may not be true fractures. However, 3 SD fractures by methods of McCloskey and Jiang, both of which require extra measures to reduce false
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4 positives, are predictive of future fractures (23). The utility of ABQ method may be best for mild fractures where it can help differentiate true osteoporotic fractures from non-fracture deformities.
Chou and Vokes associated with low BMD (31). Distinguishing short vertebral height from mild fractures is a limitation for morphometry in general and requires follow-up. Most studies comparing QM and SQ approaches conclude that QM alone may not be sufficient to diagnose vertebral fractures.
Comparison of Methods The lack of a gold standard in diagnosing vertebral fractures makes comparison of these methods difficult. The performance of Melton, Eastell, and McCloskey methods were evaluated in a study of 39 postmenopausal osteoporotic women with at least 1 osteoporotic fracture (29). The reference range for all methods was derived from 84 healthy postmenopausal women. Methods of Melton, Eastell, and McCloskey had similarly high intraobserver (range 88%e93%) and interobserver agreement (89%e94%). All methods were also compared to a reference standard determined by the SQ approach. For all methods, percent agreement with consensus reading was reasonable, ranging from 86% to 93%. As expected, sensitivity was the worst for the McCloskey method at 60%, but it had the highest specificity at 98%. Sensitivity was the highest for the Melton method at 93% with a specificity of 93%. QM does not perform as well in population-based studies, where the prevalence of vertebral fractures is lower. Compared to qualitative reading by an expert, methods of Melton, Eastell, Black, McCloskey, and Jiang all had agreement rates around 80% (23). In a study of more than 500 women and men enrolled for the European Vertebral Osteoporosis Study, QM (Eastell and McCloskey) was evaluated against a reference standard based on a combination of quantitative and qualitative assessment (20). The prevalence of vertebral fractures by reference standard was 8% in women and 5% in men. Sensitivity of QM ranged from 52% to 71% for women, compared to 20%e40% for men. In this study, QM missed up to 48% of vertebral fractures in women and up to 80% in men, particularly end plate fractures in the lumbar spine. For both genders, specificity was acceptable at 86%e98%. The combination of low prevalence and poor sensitivity resulted in a positive predictive value range of 32%e69% for women and 14%e38% in men. These gender differences may be due to a higher prevalence of osteoporosis among women than men. QM has also been compared with the SQ approach in 400 postmenopausal women with low bone mass and at least 1 vertebral fracture (8). With a reference standard set by joint consensus reading, the SQ approach performed better than QM (96% vs 92% agreement with consensus reading). Multiple cases of mild end plate fractures were given as examples in which the SQ approach identified fractures missed by QM (3 SD criterion). In another study of 500 postmenopausal women, up to 2/3 of mild fractures by SQ were missed by QM methods (30). On the other hand, false positives with the SQ method tend to occur when deformities from Scheuermann disease and degenerative change are accompanied by short vertebral height (21). Short vertebral height, especially in the mid-thoracic spine, is a normal variant that is not
Incident Vertebral Fractures As with prevalent fractures, the definition of incident fractures is also a subject of debate. One method of diagnosing incident fractures is to use QM to identify new fractures in previously normal vertebrae. However, in 1 study, this method had poor predictive value. Of 5395 vertebrae from 415 participants, 42 vertebrae were found to be deformed on the second radiograph but not on the first by the McCloskey method (32). On qualitative review, only 15 were new fractures, whereas the others were old or worsening fractures previously missed and false positives. False positives were minimized by adding a second stipulation of O10% change in vertebral heights ratio. Thus, to diagnose incident fractures, it is recommended to use absolute or percentage reduction from prior measurements to take advantage of patient- and vertebral levele specific baseline (33). A study of more than 7000 SOF participants compared several approaches to defining an incident vertebral fracture including a fixed percentage reduction in any vertebral height with and without adjustment for mean change, change in vertebral height in SDs below mean change, and a new deformity in a previously normal vertebra (34). Each method was validated based on clinical history, including age, spinal BMD, baseline vertebral fracture, height loss, and severe back pain. For all methods, there was a similar and significant association between the risk of incident vertebral fracture and each of the clinical correlates. Thus, the authors recommend a simple 20%e25% reduction in vertebral heights to define an incident vertebral fracture. Of note, vertebral heights have been noted to both decrease and increase between serial radiographs, but usually by less than 15% (35). The lack of standardization in defining incident vertebral fractures has practical implications when comparing the results of clinical trials that have led to the Food and Drug Administration approval of pharmacologic agents for osteoporosis. Different trials have used different cutoff criteria to define incident vertebral fractures. For example, the risedronate trial defined incident vertebral fractures quantitatively as loss of 15% in vertebral height that was normal at baseline and semiquantitatively as a change from grade 0 to 1e3 (36). In this study, treatment with risedronate for 3 yr resulted in a 41% risk reduction for new vertebral fractures (11% of treatment group vs 16% of placebo group). The trial evaluating alendronate had stricter criteria of a decrease of 20% and 4-mm height loss (37) and reported a 47% risk reduction in new vertebral fractures over 3 yr (8% vs 15%). Finally, the Health Outcomes and Reduced Incidence with Zoledronic Acid Once Yearly (HORIZON) trial of zoledronic acid had the same morphometric criteria as the alendronate trial but also required an increase of at least 1 SQ grade (38) and found
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Vertebral morphometry a 70% decrease in the risk of incident vertebral fractures over 3 yr (3.3% vs 10.9%). Thus, the use of various definitions of incident vertebral fractures may result in different study conclusions.
Clinical Use of QM There is no true gold standard for the diagnosis of vertebral fractures, and both qualitative and quantitative methods have been validated by correlations with BMD (20,39,40) and other clinical characteristics (30,41,42) and ability to predict future fractures (43e45). Thus, the decision to use QM is based on other factors. Compared with purely qualitative approaches, the standardization, objective criteria, and time efficiency of QM lend themselves well to use in large population-based studies. However, the placement of the 6 points is still subjective and is often performed manually (30, 46). Currently, software programs exist that place the morphometry points based on landmarks and edge detection algorithms (e.g., MorphoXpress from P&G Pharmaceuticals, SpineAnalyzer from Optasia Medical, and VFA software on both Hologic and GE densitometers). To avoid false-positive and false-negative results, the operator should correct point placement manually, if needed. These programs have been shown to save time (47) and have good reproducibility (48). In the past decade, there has been a trend to use SQ assessment either in addition or in place of QM. The combined use of SQ and QM can save time by reducing the number of radiographs requiring QM. In SOF, which enrolled more than 8000 women, radiographs were first triaged visually by trained research assistants, and QM was then performed on images flagged as probably fractured with SQ grade 1 (49). Interventional trials have also increasingly adopted the use of SQ assessment. As recommended by Genant et al in 1996 (33), the HORIZON trial with zoledronic acid defined prevalent fractures by a rather sensitive QM criteria of 3 SD that were then confirmed semiquantitatively (38). The Fracture Reduction Evaluation of Denosumab in Osteoporosis Every 6 Months (FREEDOM) trial with denosumab defined prevalent and incident fractures by SQ assessment alone (50). For clinical practice, QM would be difficult to implement as representative reference population and statistical calculations are generally required for comparison. ISCD has recommended the SQ approach for VFA because of its simplicity and additional benefit of visual inspection (51). Addition of ABQ approach would be useful particularly for mild (grade 1) fractures to distinguish true osteoporotic fractures from non-fracture deformities. In conclusion, like many diagnostic criteria, QM sets a rather arbitrary cutoff in a spectrum of vertebral deformities. Also, not all deformities are due to fractures, which can be discerned by qualitative approaches. However, the use of various methods of QM has been consistent in determining the prevalence of vertebral fractures in population-based studies. More importantly, from a clinical perspective, treatment of osteoporosis has been shown to decrease the risk
5 of incident morphometric vertebral fractures, which are associated with increased mortality risk. Thus, QM remains one of the options for use in large epidemiological and interventional trials as it provides a standard, reasonably reproducible approach. However, it is best to combine it with some form of visual assessment to minimize the false positive and negatives resulting from non-fracture deformities and poor sensitivity of QM in detecting more subtle end plate fractures.
References 1. Genant HK, Wu CY, van Kuijk C, et al. 1993 Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 9:1137e1148. 2. de Bruijne M, Lund MT, Tanko LB, et al. 2007 Quantitative vertebral morphometry using neighbor-conditional shape models. Med Image Anal 5:503e512. 3. Smyth PP, Taylor CJ, Adams JE. 1999 Vertebral shape: automatic measurement with active shape models. Radiology 2: 571e578. 4. Stern D, Njagulj V, Likar B, et al. 2013 Quantitative vertebral morphometry based on parametric modeling of vertebral bodies in 3D. Osteoporos Int 4:1357e1368. 5. Grigoryan M, Guermazi A, Roemer FW, et al. 2003 Recognizing and reporting osteoporotic vertebral fractures. Eur Spine J 12 Suppl 2:S104eS112. 6. Hurxthal LM. 1968 Measurement of anterior vertebral compressions and biconcave vertebrae. Am J Roentgenol 3: 635e644. 7. Edmondston SJ, Price RI, Valente B, Singer KP. 1999 Measurement of vertebral body heights: ex vivo comparisons between morphometric X-ray absorptiometry, morphometric radiography and direct measurements. Osteoporos Int 1:7e13. 8. Wu CY, Li J, Jergas M, Genant HK. 1995 Comparison of semiquantitative and quantitative techniques for the assessment of prevalent and incident vertebral fractures. Osteoporos Int 5: 354e370. 9. Melton LJ 3rd, Kan SH, Frye MA, et al. 1989 Epidemiology of vertebral fractures in women. Am J Epidemiol 5:1000e1011. 10. Eastell R, Cedel SL, Wahner HW, et al. 1991 Classification of vertebral fractures. J Bone Miner Res 3:207e215. 11. McCloskey EV, Spector TD, Eyres KS, et al. 1993 The assessment of vertebral deformity: a method for use in population studies and clinical trials. Osteoporos Int 3:138e147. 12. Black DM, Cummings SR, Stone K, et al. 1991 A new approach to defining normal vertebral dimensions. J Bone Miner Res 8:883e892. 13. Zebaze RM, Maalouf G, Wehbe J, et al. 2004 The varying distribution of intra- and inter-vertebral height ratios determines the prevalence of vertebral fractures. Bone 2:348e356. 14. Jiang G, Ferrar L, Barrington NA, Eastell R. 2007 Standardised quantitative morphometry: a modified approach for quantitative identification of prevalent vertebral deformities. Osteoporos Int 10:1411e1419. 15. Minne HW, Leidig G, Wuster C, et al. 1988 A newly developed spine deformity index (SDI) to quantitate vertebral crush fractures in patients with osteoporosis. Bone Miner 4:335e349. 16. O’Neill TW, Varlow J, Felsenberg D, et al. 1994 Variation in vertebral height ratios in population studies. European Vertebral Osteoporosis Study Group. J Bone Miner Res 12:1895e 1907. 17. Grados F, Fardellone P, Benammar M, et al. 1999 Influence of age and sex on vertebral shape indices assessed by radiographic morphometry. Osteoporos Int 6:450e455.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6 18. Diacinti D, Pisani D, Del Fiacco R, et al. 2011 Vertebral morphometry by X-ray absorptiometry: which reference data for vertebral heights? Bone 3:526e536. 19. Rea JA, Steiger P, Blake GM, et al. 1998 Morphometric X-ray absorptiometry: reference data for vertebral dimensions. J Bone Miner Res 3:464e474. 20. Leidig-Bruckner G, Limberg B, Felsenberg D, et al. 2000 Sex difference in the validity of vertebral deformities as an index of prevalent vertebral osteoporotic fractures: a population survey of older men and women. Osteoporos Int 2:102e119. 21. Jiang G, Eastell R, Barrington NA, Ferrar L. 2004 Comparison of methods for the visual identification of prevalent vertebral fracture in osteoporosis. Osteoporos Int 11:887e896. 22. Ferrar L, Jiang G, Cawthon PM, et al. 2007 Identification of vertebral fracture and non-osteoporotic short vertebral height in men: the MrOS study. J Bone Miner Res 9: 1434e1441. 23. Melton LJ 3rd, Wenger DE, Atkinson EJ, et al. 2006 Influence of baseline deformity definition on subsequent vertebral fracture risk in postmenopausal women. Osteoporos Int 7: 978e985. 24. O’Neill TW, Felsenberg D, Varlow J, et al. 1996 The prevalence of vertebral deformity in European men and women: the European Vertebral Osteoporosis Study. J Bone Miner Res 7:1010e1018. 25. Jackson SA, Tenenhouse A, Robertson L. 2000 Vertebral fracture definition from population-based data: preliminary results from the Canadian Multicenter Osteoporosis Study (CaMos). Osteoporos Int 8:680e687. 26. Ettinger B, Black DM, Nevitt MC, et al. 1992 Contribution of vertebral deformities to chronic back pain and disability. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 4:449e456. 27. Burger H, Van Daele PL, Grashuis K, et al. 1997 Vertebral deformities and functional impairment in men and women. J Bone Miner Res 1:152e157. 28. Jones G, White C, Nguyen T, et al. 1996 Prevalent vertebral deformities: relationship to bone mineral density and spinal osteophytosis in elderly men and women. Osteoporos Int 3: 233e239. 29. Grados F, Roux C, de Vernejoul MC, et al. 2001 Comparison of four morphometric definitions and a semiquantitative consensus reading for assessing prevalent vertebral fractures. Osteoporos Int 9:716e722. 30. Black DM, Palermo L, Nevitt MC, et al. 1995 Comparison of methods for defining prevalent vertebral deformities: the Study of Osteoporotic Fractures. J Bone Miner Res 6:890e902. 31. Ferrar L, Jiang G, Armbrecht G, et al. 2007 Is short vertebral height always an osteoporotic fracture? The Osteoporosis and Ultrasound Study (OPUS). Bone 1:5e12. 32. Weber K, Lunt M, Gowin W, et al. 1999 Measurement imprecision in vertebral morphometry of spinal radiographs obtained in the European Prospective Osteoporosis Study: consequences for the identification of prevalent and incident deformities. Br J Radiol 862:957e966. 33. Genant HK, Jergas M, Palermo L, et al. 1996 Comparison of semiquantitative visual and quantitative morphometric assessment of prevalent and incident vertebral fractures in osteoporosis The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 7:984e996.
Chou and Vokes 34. Black DM, Palermo L, Nevitt MC, et al. 1999 Defining incident vertebral deformity: a prospective comparison of several approaches. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 1:90e101. 35. Nelson DA, Kleerekoper M, Peterson EL. 1994 Reversal of vertebral deformities in osteoporosis: measurement error or ‘‘rebound’’? J Bone Miner Res 7:977e982. 36. Harris ST, Watts NB, Genant HK, et al. 1999 Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis: a randomized controlled trial. Vertebral Efficacy With Risedronate Therapy (VERT) Study Group. JAMA 14:1344e1352. 37. Black DM, Cummings SR, Karpf DB, et al. 1996 Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Fracture Intervention Trial Research Group. Lancet 9041:1535e1541. 38. Black DM, Delmas PD, Eastell R, et al. 2007 Once-yearly zoledronic acid for treatment of postmenopausal osteoporosis. N Engl J Med 18:1809e1822. 39. Mann T, Oviatt SK, Wilson D, et al. 1992 Vertebral deformity in men. J Bone Miner Res 11:1259e1265. 40. Cauley JA, Zmuda JM, Wisniewski SR, et al. 2004 Bone mineral density and prevalent vertebral fractures in men and women. Osteoporos Int 1:32e37. 41. Jinbayashi H, Aoyagi K, Ross PD, et al. 2002 Prevalence of vertebral deformity and its associations with physical impairment among Japanese women: the Hizen-Oshima Study. Osteoporos Int 9:723e730. 42. Hasserius R, Johnell O, Nilsson BE, et al. 2003 Hip fracture patients have more vertebral deformities than subjects in population-based studies. Bone 2:180e184. 43. Ross PD, Davis JW, Epstein RS, Wasnich RD. 1991 Preexisting fractures and bone mass predict vertebral fracture incidence in women. Ann Intern Med 11:919e923. 44. Black DM, Arden NK, Palermo L, et al. 1999 Prevalent vertebral deformities predict hip fractures and new vertebral deformities but not wrist fractures. Study of Osteoporotic Fractures Research Group. J Bone Miner Res 5:821e828. 45. Cauley JA, Hochberg MC, Lui LY, et al. 2007 Long-term risk of incident vertebral fractures. JAMA 23:2761e2767. 46. Hasserius R, Redlund-Johnell I, Mellstrom D, et al. 2001 Vertebral deformation in urban Swedish men and women: prevalence based on 797 subjects. Acta Orthop Scand 3:273e278. 47. Guglielmi G, Palmieri F, Placentino MG, et al. 2009 Assessment of osteoporotic vertebral fractures using specialized workflow software for 6-point morphometry. Eur J Radiol 1: 142e148. 48. Guglielmi G, Stoppino LP, Placentino MG, et al. 2009 Reproducibility of a semi-automatic method for 6-point vertebral morphometry in a multi-centre trial. Eur J Radiol 1:173e178. 49. Cauley JA, Palermo L, Vogt M, et al. 2008 Prevalent vertebral fractures in black women and white women. J Bone Miner Res 9:1458e1467. 50. Cummings SR, San Martin J, McClung MR, et al. 2009 Denosumab for prevention of fractures in postmenopausal women with osteoporosis. N Engl J Med 8:756e765. 51. ISCD. 2013 ISCD official positionsdAdult. Available at: http://www.iscd.org/official-positions/2013-iscd-official-positi ons-adult/. Accessed September 18, 2014.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Original Article
Vertebral Morphometry Sharon H. Chou,1 and Tamara Vokes*,2 1
Division of Endocrinology, Diabetes, and Hypertension, Brigham and Women’s Hospital, Boston, MA, USA; and 2Section of Adult & Pediatric Endocrinology, Diabetes, Metabolism, The University of Chicago, Chicago, IL, USA
Abstract There is as yet no agreement about the criteria by which to arrive at an imaging diagnosis of a vertebral fracture. Because high-grade fractures result in alterations in vertebral shape, 1 possible avenue of diagnosis has been to quantitate changes in vertebral shape. The result has been a variety of methods for the relative and absolute measurements of vertebral dimensions. Such measurements have also lent themselves to automated computed analysis. The number of techniques reflects the absence of any consensus about the best. The semiquantitative technique proposed by Genant has become the most widely used and has served the field well for comparative purposes. Its use in higher grade fractures has been widely endorsed, if some concepts (e.g., short vertebral heightevertebrae) are at variance with lower grades of fracturing. Vertebral morphometry may be the only recourse in high volume epidemiological and interventional studies. Key Words: Vertebral fractures; quantitative morphometry; semiquantitative morphometry.
Quantitative morphometry (QM) methods involve actual measurement of vertebral dimensions with the most commonly used method being 6-point morphometry. In addition to the latter, there are other methods based on the shape of the vertebral body (2,3). More recently, QM by parametric modeling of vertebral bodies in 3 dimensions has been developed (4). In this article, we will focus on 6-point morphometry as most of reported studies have used this method.
Introduction The diagnosis of vertebral fractures is often based on changes in shape due to a compressive force along the superioreinferior axis resulting in a fracture of the vertebral end plate. However, due to the nature of the changes, there is no consensus about the definition or grading of fractures. Broadly, the methods for diagnosing vertebral fractures can be classified as qualitative, quantitative, and semiquantitative (SQ). Qualitatively, a radiologist or other trained observer inspects the spine image(s) and decides whether a given vertebra is fractured. The diagnosis of vertebral fracture by this method is based on several factors such as changes in shape including deformity of the end plate, buckling of cortices, and loss of vertebral height. SQ methods, of which the method of Genant et al (1) has been the most widely used, rely on visual identification of abnormal vertebral shape and comparison of its appearance to a chart (Fig. 1).
Technique: Six-Point Morphometry On lateral spine images (radiographs or dual-energy X-ray absorptiometry spine images, termed vertebral fracture assessment [VFA]), points are manually or digitally placed on each of the 4 corners of the vertebral body, and 2 additional points are placed in the center of the upper and lower end plates (Fig. 2A). These 6 points are used to determine the anterior, mid, and posterior vertebral heights of each vertebra (Fig. 2A). Using these points, each vertebra is then assigned a type and grade (severity) of fracture. The type of deformity is based on lower than expected ratio of particular vertebral heights: wedge for decreased ratio of anterior-toposterior heights (Fig. 3A), biconcave for decreased ratio of mid-to-posterior heights, and compression (or crush) for decreased ratio of posterior height to the posterior height of
Received 10/24/14; Accepted 08/12/15. *Address correspondence to: Tamara Vokes, MD, Section of Adult & Pediatric Endocrinology, Diabetes, Metabolism, The University of Chicago, 5841 South Maryland Avenue, MC 1027, Chicago, IL 60637. E-mail: [email protected]
1
2
Chou and Vokes
Fig. 1. Genant semiquantitative scale.
adjacent vertebrae (Fig. 3B). However, there is normal variation in the shape of vertebral bodies: mid-thoracic vertebrae may be slightly wedge shaped, and normal lumbar vertebrae may have a biconcave shape (5). How to define what degree of deformity constitutes a fracture and how to assign it a grade are a matter of debate with several methods studied (see in the following section) but no consensus reached. Although QM is intended to be quantitative, point placement is subjective and not always straightforward. Osteophytes, Schmorl nodes, and the uncinate process at the posterosuperior border of the thoracic vertebrae should be excluded for point placement (6; Fig. 4A). Point placement may be challenging for images with these non-fracture deformities and for those with projectional variations resulting from radiographic technique (Figs. 2B and 4B). Despite subjectivity and difficulty with point placement, the accuracy and precision of QM are reasonably good for well-trained observers. Accuracy was assessed in a small ex vivo study using 9 cadaveric vertebral columns (7). Anterior, mid, and posterior vertebral heights were assessed by direct measurement and by QM performed on radiographs, as well as on dual-energy X-ray absorptiometry images. Both QM approaches were strongly correlated (r2 5 0.99) with direct measurements. Reproducibility of QM was also found to be good in 400 postmenopausal women with low
Fig. 2. Placement of morphometric points in a vertebra with orthogonal projection (A) and one with obliquity (B). Vertebral heights: anterior (a), mid-vertebral (m), and posterior (p). From http://www.advances-in-medicine.net/2011/ 12/conventional-x-rays-in-the-diagnosis-of-osteoporosis-morp hometric-vertebral-fracture-analysis.html.
Fig. 3. Examples of spine radiographs with fractures as indicated by the white arrows: wedge fracture at T7danterior height is lower than posterior height (A); crush fracture at T5dposterior height of T5 is lower than the posterior heights of T4 and T6 (B). Crush fractures typically have decreased vertebral heights throughout the vertebra. Please note that point placement can be debated in these examples, exemplifying some of the uncertainties in vertebral morphometry. bone mass and at least 1 vertebral fracture with coefficients of variation of 2%e4% for both intraobserver and interobserver comparisons of vertebral heights (8).
Defining Prevalent Vertebral Fractures There is no consensus about defining vertebral fractures by QM. Vertebral fractures can be defined as a reduction in vertebral height compared with other heights within the vertebra or to a normal height as determined by a reference population. Furthermore, height differences that constitute a fracture can be based on percent reduction or standard deviation (SD) from the norm. The Melton Method defined a vertebral fracture as any ratio of anterior-to-posterior height, mid-to-posterior height, or posterior-to-posterior height of adjacent vertebrae that is less than 0.85 (9). To account for natural variations in vertebral shape and size at each level, the individual’s ratio may be
Fig. 4. Challenges to point placement on spine radiographs: avoidance of labeling osteophytes (A) and difficulty (subjectivity) of point placement with obliquity (B). Please note that point placement in these examples are debatable.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Vertebral morphometry adjusted by a vertebra-specific factor that normalizes it to a reference population without vertebral fractures. As expected, when the ratios were adjusted, the prevalence of vertebral fractures in 200 women aged 50 yr decreased from 83% to 26%, closer to the prevalence of 28% as determined by qualitative reading. The Eastell Method defined fracture simply as a ratio of vertebral heights that is more than 3 SD below the normal mean for that vertebral level (10). The normal mean was derived from 52 healthy women without vertebral fractures. Fractures of 3e4 SD are classified as grade 1, and fractures greater than 4 SDs are grade 2. The McCloskey Method aimed to limit the number of false-positive results with the Eastell Method by also requiring that the ratio of any measured vertebral height to predicted posterior height based on adjacent normal vertebrae must also be greater than 3 SD below the mean (11). The Black Method modified the method of Eastell by using a reference population of 2992 women aged 65e70 yr from the Study of Osteoporotic Fractures (SOF) that included women with vertebral fractures but excluded the outermost data by statistical trimming (12). Fractures were defined as values greater than 3 SD below the mean. This method assumes that the ratios among non-fractured vertebral bodies obeys a Gaussian or normal distribution, which may not be the case (13), and the prevalence of abnormal vertebrae at any level is !10%. Other similar methods have been described by Jiang et al (14), Minne et al (15), and others. Most of these quantitative methods require comparison to a reference population. This population should be similar to the study population in terms of age and demographics (16e18). Separate reference values for men and women should also be used (16). The reference range can be derived from a similar population with or without vertebral deformities. If the reference population contains some women (!30%) with vertebral deformities, the statistical methods described by Black et al (12) should be considered. Statistical trimming of a reference population with high prevalence of vertebral fractures, however, will increase false negatives (14). Finally, the source of the image (conventional spinal radiographs vs VFA images) should also be consistent between the study and reference populations (19).
The Genant Semiquantitative Method Because QM is strictly based on vertebral measurements, it may misclassify as fractures some non-fracture deformities such as those resulting from Scheuermann disease or osteoarthritis (20). To avoid such errors and simplify the diagnosis of vertebral fractures, Genant et al (1) introduced an SQ method. With this approach, each vertebra from T4 to L4 is assessed visually for height loss. Signs of fracture are described and include lack of end plate parallelism, end plate depression, buckling of cortical margin, and loss of vertical continuity with adjacent vertebrae, but these are not required for diagnosis. Once a vertebra is designated as fractured, then
3 the grade is assigned by visually comparing it to a chart (Fig. 1): height loss of approximately 20%e25% is labeled as grade 1 (mild), 25%e40% as grade 2 (moderate), and O40% as grade 3 (severe) compared with other heights of the same and/or neighboring vertebrae. Unlike quantitative methods, this approach does not require a reference population for comparison. However, this approach does require training to distinguish fracture from non-fracture deformity or normal variant and suffers from overlap with the short vertebral height variant (21). Despite this, intraobserver variability has been found to be low with percentage agreement between repeat readings of 93% for inexperienced and 97% for experienced observers (1). Interobserver agreement is similarly high at 94% (1,8).
Algorithm-Based Qualitative Assessment The algorithm-based qualitative (ABQ) method is a more recent development resulting from the possible shortcomings of QM and SQ method of Genant (21). Instead of focusing on reductions in vertebral height, the ABQ method focuses on the presence of end plate deformity. The diagnosis of an osteoporotic vertebral fracture by the ABQ method requires an end plate depression, which is symmetrical, concave, involving the whole end plate depression within the rim, and is not associated with trauma, tumor, or metabolic disease. Vertebrae with end plate depressions not meeting additional criteria or vertebrae with no end plate fracture but short vertebral height are classified as non-fracture deformities, developmental variant, non-osteoporotic fracture, or other condition. Prevalence of vertebral fractures in postmenopausal women is lower with ABQ than SQ methods (7% vs 24%) (21) but was not much different in older men (10% vs 13%; 22).
Mild Vs Severe Fractures Mild fractures are controversial in terms of diagnosis and clinical implications. Mild fractures are deformities resulting in a height reduction to 3e4 SD below the mean (about 20% height loss), and more severe fractures are 4 SD below the mean. Depending on the cutoff criteria (and reference population), the prevalence of vertebral fractures can range from 7% to 90% (23). In general, when including mild fractures (3 SD and 15% reduction), the prevalence of vertebral fractures in postmenopausal women is about 20%e25% (9,10,12,24,25). More stringent criteria (4 SD) decrease the prevalence to about 10% (10,25). Unlike 4 SD fractures, 3 SD vertebral fractures are not associated with increased back pain, disability, height loss, or lower lumbar spine bone mineral density (BMD) (20,26e28). Furthermore, 3 SD fractures as determined by methods by Eastell and Black and fractures of 15%e20% reduction in height are not predictive of future vertebral fractures (23). These deformities may not be true fractures. However, 3 SD fractures by methods of McCloskey and Jiang, both of which require extra measures to reduce false
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4 positives, are predictive of future fractures (23). The utility of ABQ method may be best for mild fractures where it can help differentiate true osteoporotic fractures from non-fracture deformities.
Chou and Vokes associated with low BMD (31). Distinguishing short vertebral height from mild fractures is a limitation for morphometry in general and requires follow-up. Most studies comparing QM and SQ approaches conclude that QM alone may not be sufficient to diagnose vertebral fractures.
Comparison of Methods The lack of a gold standard in diagnosing vertebral fractures makes comparison of these methods difficult. The performance of Melton, Eastell, and McCloskey methods were evaluated in a study of 39 postmenopausal osteoporotic women with at least 1 osteoporotic fracture (29). The reference range for all methods was derived from 84 healthy postmenopausal women. Methods of Melton, Eastell, and McCloskey had similarly high intraobserver (range 88%e93%) and interobserver agreement (89%e94%). All methods were also compared to a reference standard determined by the SQ approach. For all methods, percent agreement with consensus reading was reasonable, ranging from 86% to 93%. As expected, sensitivity was the worst for the McCloskey method at 60%, but it had the highest specificity at 98%. Sensitivity was the highest for the Melton method at 93% with a specificity of 93%. QM does not perform as well in population-based studies, where the prevalence of vertebral fractures is lower. Compared to qualitative reading by an expert, methods of Melton, Eastell, Black, McCloskey, and Jiang all had agreement rates around 80% (23). In a study of more than 500 women and men enrolled for the European Vertebral Osteoporosis Study, QM (Eastell and McCloskey) was evaluated against a reference standard based on a combination of quantitative and qualitative assessment (20). The prevalence of vertebral fractures by reference standard was 8% in women and 5% in men. Sensitivity of QM ranged from 52% to 71% for women, compared to 20%e40% for men. In this study, QM missed up to 48% of vertebral fractures in women and up to 80% in men, particularly end plate fractures in the lumbar spine. For both genders, specificity was acceptable at 86%e98%. The combination of low prevalence and poor sensitivity resulted in a positive predictive value range of 32%e69% for women and 14%e38% in men. These gender differences may be due to a higher prevalence of osteoporosis among women than men. QM has also been compared with the SQ approach in 400 postmenopausal women with low bone mass and at least 1 vertebral fracture (8). With a reference standard set by joint consensus reading, the SQ approach performed better than QM (96% vs 92% agreement with consensus reading). Multiple cases of mild end plate fractures were given as examples in which the SQ approach identified fractures missed by QM (3 SD criterion). In another study of 500 postmenopausal women, up to 2/3 of mild fractures by SQ were missed by QM methods (30). On the other hand, false positives with the SQ method tend to occur when deformities from Scheuermann disease and degenerative change are accompanied by short vertebral height (21). Short vertebral height, especially in the mid-thoracic spine, is a normal variant that is not
Incident Vertebral Fractures As with prevalent fractures, the definition of incident fractures is also a subject of debate. One method of diagnosing incident fractures is to use QM to identify new fractures in previously normal vertebrae. However, in 1 study, this method had poor predictive value. Of 5395 vertebrae from 415 participants, 42 vertebrae were found to be deformed on the second radiograph but not on the first by the McCloskey method (32). On qualitative review, only 15 were new fractures, whereas the others were old or worsening fractures previously missed and false positives. False positives were minimized by adding a second stipulation of O10% change in vertebral heights ratio. Thus, to diagnose incident fractures, it is recommended to use absolute or percentage reduction from prior measurements to take advantage of patient- and vertebral levele specific baseline (33). A study of more than 7000 SOF participants compared several approaches to defining an incident vertebral fracture including a fixed percentage reduction in any vertebral height with and without adjustment for mean change, change in vertebral height in SDs below mean change, and a new deformity in a previously normal vertebra (34). Each method was validated based on clinical history, including age, spinal BMD, baseline vertebral fracture, height loss, and severe back pain. For all methods, there was a similar and significant association between the risk of incident vertebral fracture and each of the clinical correlates. Thus, the authors recommend a simple 20%e25% reduction in vertebral heights to define an incident vertebral fracture. Of note, vertebral heights have been noted to both decrease and increase between serial radiographs, but usually by less than 15% (35). The lack of standardization in defining incident vertebral fractures has practical implications when comparing the results of clinical trials that have led to the Food and Drug Administration approval of pharmacologic agents for osteoporosis. Different trials have used different cutoff criteria to define incident vertebral fractures. For example, the risedronate trial defined incident vertebral fractures quantitatively as loss of 15% in vertebral height that was normal at baseline and semiquantitatively as a change from grade 0 to 1e3 (36). In this study, treatment with risedronate for 3 yr resulted in a 41% risk reduction for new vertebral fractures (11% of treatment group vs 16% of placebo group). The trial evaluating alendronate had stricter criteria of a decrease of 20% and 4-mm height loss (37) and reported a 47% risk reduction in new vertebral fractures over 3 yr (8% vs 15%). Finally, the Health Outcomes and Reduced Incidence with Zoledronic Acid Once Yearly (HORIZON) trial of zoledronic acid had the same morphometric criteria as the alendronate trial but also required an increase of at least 1 SQ grade (38) and found
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Vertebral morphometry a 70% decrease in the risk of incident vertebral fractures over 3 yr (3.3% vs 10.9%). Thus, the use of various definitions of incident vertebral fractures may result in different study conclusions.
Clinical Use of QM There is no true gold standard for the diagnosis of vertebral fractures, and both qualitative and quantitative methods have been validated by correlations with BMD (20,39,40) and other clinical characteristics (30,41,42) and ability to predict future fractures (43e45). Thus, the decision to use QM is based on other factors. Compared with purely qualitative approaches, the standardization, objective criteria, and time efficiency of QM lend themselves well to use in large population-based studies. However, the placement of the 6 points is still subjective and is often performed manually (30, 46). Currently, software programs exist that place the morphometry points based on landmarks and edge detection algorithms (e.g., MorphoXpress from P&G Pharmaceuticals, SpineAnalyzer from Optasia Medical, and VFA software on both Hologic and GE densitometers). To avoid false-positive and false-negative results, the operator should correct point placement manually, if needed. These programs have been shown to save time (47) and have good reproducibility (48). In the past decade, there has been a trend to use SQ assessment either in addition or in place of QM. The combined use of SQ and QM can save time by reducing the number of radiographs requiring QM. In SOF, which enrolled more than 8000 women, radiographs were first triaged visually by trained research assistants, and QM was then performed on images flagged as probably fractured with SQ grade 1 (49). Interventional trials have also increasingly adopted the use of SQ assessment. As recommended by Genant et al in 1996 (33), the HORIZON trial with zoledronic acid defined prevalent fractures by a rather sensitive QM criteria of 3 SD that were then confirmed semiquantitatively (38). The Fracture Reduction Evaluation of Denosumab in Osteoporosis Every 6 Months (FREEDOM) trial with denosumab defined prevalent and incident fractures by SQ assessment alone (50). For clinical practice, QM would be difficult to implement as representative reference population and statistical calculations are generally required for comparison. ISCD has recommended the SQ approach for VFA because of its simplicity and additional benefit of visual inspection (51). Addition of ABQ approach would be useful particularly for mild (grade 1) fractures to distinguish true osteoporotic fractures from non-fracture deformities. In conclusion, like many diagnostic criteria, QM sets a rather arbitrary cutoff in a spectrum of vertebral deformities. Also, not all deformities are due to fractures, which can be discerned by qualitative approaches. However, the use of various methods of QM has been consistent in determining the prevalence of vertebral fractures in population-based studies. More importantly, from a clinical perspective, treatment of osteoporosis has been shown to decrease the risk
5 of incident morphometric vertebral fractures, which are associated with increased mortality risk. Thus, QM remains one of the options for use in large epidemiological and interventional trials as it provides a standard, reasonably reproducible approach. However, it is best to combine it with some form of visual assessment to minimize the false positive and negatives resulting from non-fracture deformities and poor sensitivity of QM in detecting more subtle end plate fractures.
References 1. Genant HK, Wu CY, van Kuijk C, et al. 1993 Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 9:1137e1148. 2. de Bruijne M, Lund MT, Tanko LB, et al. 2007 Quantitative vertebral morphometry using neighbor-conditional shape models. Med Image Anal 5:503e512. 3. Smyth PP, Taylor CJ, Adams JE. 1999 Vertebral shape: automatic measurement with active shape models. Radiology 2: 571e578. 4. Stern D, Njagulj V, Likar B, et al. 2013 Quantitative vertebral morphometry based on parametric modeling of vertebral bodies in 3D. Osteoporos Int 4:1357e1368. 5. Grigoryan M, Guermazi A, Roemer FW, et al. 2003 Recognizing and reporting osteoporotic vertebral fractures. Eur Spine J 12 Suppl 2:S104eS112. 6. Hurxthal LM. 1968 Measurement of anterior vertebral compressions and biconcave vertebrae. Am J Roentgenol 3: 635e644. 7. Edmondston SJ, Price RI, Valente B, Singer KP. 1999 Measurement of vertebral body heights: ex vivo comparisons between morphometric X-ray absorptiometry, morphometric radiography and direct measurements. Osteoporos Int 1:7e13. 8. Wu CY, Li J, Jergas M, Genant HK. 1995 Comparison of semiquantitative and quantitative techniques for the assessment of prevalent and incident vertebral fractures. Osteoporos Int 5: 354e370. 9. Melton LJ 3rd, Kan SH, Frye MA, et al. 1989 Epidemiology of vertebral fractures in women. Am J Epidemiol 5:1000e1011. 10. Eastell R, Cedel SL, Wahner HW, et al. 1991 Classification of vertebral fractures. J Bone Miner Res 3:207e215. 11. McCloskey EV, Spector TD, Eyres KS, et al. 1993 The assessment of vertebral deformity: a method for use in population studies and clinical trials. Osteoporos Int 3:138e147. 12. Black DM, Cummings SR, Stone K, et al. 1991 A new approach to defining normal vertebral dimensions. J Bone Miner Res 8:883e892. 13. Zebaze RM, Maalouf G, Wehbe J, et al. 2004 The varying distribution of intra- and inter-vertebral height ratios determines the prevalence of vertebral fractures. Bone 2:348e356. 14. Jiang G, Ferrar L, Barrington NA, Eastell R. 2007 Standardised quantitative morphometry: a modified approach for quantitative identification of prevalent vertebral deformities. Osteoporos Int 10:1411e1419. 15. Minne HW, Leidig G, Wuster C, et al. 1988 A newly developed spine deformity index (SDI) to quantitate vertebral crush fractures in patients with osteoporosis. Bone Miner 4:335e349. 16. O’Neill TW, Varlow J, Felsenberg D, et al. 1994 Variation in vertebral height ratios in population studies. European Vertebral Osteoporosis Study Group. J Bone Miner Res 12:1895e 1907. 17. Grados F, Fardellone P, Benammar M, et al. 1999 Influence of age and sex on vertebral shape indices assessed by radiographic morphometry. Osteoporos Int 6:450e455.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6 18. Diacinti D, Pisani D, Del Fiacco R, et al. 2011 Vertebral morphometry by X-ray absorptiometry: which reference data for vertebral heights? Bone 3:526e536. 19. Rea JA, Steiger P, Blake GM, et al. 1998 Morphometric X-ray absorptiometry: reference data for vertebral dimensions. J Bone Miner Res 3:464e474. 20. Leidig-Bruckner G, Limberg B, Felsenberg D, et al. 2000 Sex difference in the validity of vertebral deformities as an index of prevalent vertebral osteoporotic fractures: a population survey of older men and women. Osteoporos Int 2:102e119. 21. Jiang G, Eastell R, Barrington NA, Ferrar L. 2004 Comparison of methods for the visual identification of prevalent vertebral fracture in osteoporosis. Osteoporos Int 11:887e896. 22. Ferrar L, Jiang G, Cawthon PM, et al. 2007 Identification of vertebral fracture and non-osteoporotic short vertebral height in men: the MrOS study. J Bone Miner Res 9: 1434e1441. 23. Melton LJ 3rd, Wenger DE, Atkinson EJ, et al. 2006 Influence of baseline deformity definition on subsequent vertebral fracture risk in postmenopausal women. Osteoporos Int 7: 978e985. 24. O’Neill TW, Felsenberg D, Varlow J, et al. 1996 The prevalence of vertebral deformity in European men and women: the European Vertebral Osteoporosis Study. J Bone Miner Res 7:1010e1018. 25. Jackson SA, Tenenhouse A, Robertson L. 2000 Vertebral fracture definition from population-based data: preliminary results from the Canadian Multicenter Osteoporosis Study (CaMos). Osteoporos Int 8:680e687. 26. Ettinger B, Black DM, Nevitt MC, et al. 1992 Contribution of vertebral deformities to chronic back pain and disability. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 4:449e456. 27. Burger H, Van Daele PL, Grashuis K, et al. 1997 Vertebral deformities and functional impairment in men and women. J Bone Miner Res 1:152e157. 28. Jones G, White C, Nguyen T, et al. 1996 Prevalent vertebral deformities: relationship to bone mineral density and spinal osteophytosis in elderly men and women. Osteoporos Int 3: 233e239. 29. Grados F, Roux C, de Vernejoul MC, et al. 2001 Comparison of four morphometric definitions and a semiquantitative consensus reading for assessing prevalent vertebral fractures. Osteoporos Int 9:716e722. 30. Black DM, Palermo L, Nevitt MC, et al. 1995 Comparison of methods for defining prevalent vertebral deformities: the Study of Osteoporotic Fractures. J Bone Miner Res 6:890e902. 31. Ferrar L, Jiang G, Armbrecht G, et al. 2007 Is short vertebral height always an osteoporotic fracture? The Osteoporosis and Ultrasound Study (OPUS). Bone 1:5e12. 32. Weber K, Lunt M, Gowin W, et al. 1999 Measurement imprecision in vertebral morphometry of spinal radiographs obtained in the European Prospective Osteoporosis Study: consequences for the identification of prevalent and incident deformities. Br J Radiol 862:957e966. 33. Genant HK, Jergas M, Palermo L, et al. 1996 Comparison of semiquantitative visual and quantitative morphometric assessment of prevalent and incident vertebral fractures in osteoporosis The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 7:984e996.
Chou and Vokes 34. Black DM, Palermo L, Nevitt MC, et al. 1999 Defining incident vertebral deformity: a prospective comparison of several approaches. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 1:90e101. 35. Nelson DA, Kleerekoper M, Peterson EL. 1994 Reversal of vertebral deformities in osteoporosis: measurement error or ‘‘rebound’’? J Bone Miner Res 7:977e982. 36. Harris ST, Watts NB, Genant HK, et al. 1999 Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis: a randomized controlled trial. Vertebral Efficacy With Risedronate Therapy (VERT) Study Group. JAMA 14:1344e1352. 37. Black DM, Cummings SR, Karpf DB, et al. 1996 Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Fracture Intervention Trial Research Group. Lancet 9041:1535e1541. 38. Black DM, Delmas PD, Eastell R, et al. 2007 Once-yearly zoledronic acid for treatment of postmenopausal osteoporosis. N Engl J Med 18:1809e1822. 39. Mann T, Oviatt SK, Wilson D, et al. 1992 Vertebral deformity in men. J Bone Miner Res 11:1259e1265. 40. Cauley JA, Zmuda JM, Wisniewski SR, et al. 2004 Bone mineral density and prevalent vertebral fractures in men and women. Osteoporos Int 1:32e37. 41. Jinbayashi H, Aoyagi K, Ross PD, et al. 2002 Prevalence of vertebral deformity and its associations with physical impairment among Japanese women: the Hizen-Oshima Study. Osteoporos Int 9:723e730. 42. Hasserius R, Johnell O, Nilsson BE, et al. 2003 Hip fracture patients have more vertebral deformities than subjects in population-based studies. Bone 2:180e184. 43. Ross PD, Davis JW, Epstein RS, Wasnich RD. 1991 Preexisting fractures and bone mass predict vertebral fracture incidence in women. Ann Intern Med 11:919e923. 44. Black DM, Arden NK, Palermo L, et al. 1999 Prevalent vertebral deformities predict hip fractures and new vertebral deformities but not wrist fractures. Study of Osteoporotic Fractures Research Group. J Bone Miner Res 5:821e828. 45. Cauley JA, Hochberg MC, Lui LY, et al. 2007 Long-term risk of incident vertebral fractures. JAMA 23:2761e2767. 46. Hasserius R, Redlund-Johnell I, Mellstrom D, et al. 2001 Vertebral deformation in urban Swedish men and women: prevalence based on 797 subjects. Acta Orthop Scand 3:273e278. 47. Guglielmi G, Palmieri F, Placentino MG, et al. 2009 Assessment of osteoporotic vertebral fractures using specialized workflow software for 6-point morphometry. Eur J Radiol 1: 142e148. 48. Guglielmi G, Stoppino LP, Placentino MG, et al. 2009 Reproducibility of a semi-automatic method for 6-point vertebral morphometry in a multi-centre trial. Eur J Radiol 1:173e178. 49. Cauley JA, Palermo L, Vogt M, et al. 2008 Prevalent vertebral fractures in black women and white women. J Bone Miner Res 9:1458e1467. 50. Cummings SR, San Martin J, McClung MR, et al. 2009 Denosumab for prevention of fractures in postmenopausal women with osteoporosis. N Engl J Med 8:756e765. 51. ISCD. 2013 ISCD official positionsdAdult. Available at: http://www.iscd.org/official-positions/2013-iscd-official-positi ons-adult/. Accessed September 18, 2014.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015