- Email: [email protected]
Opportunistic Identification of Vertebral Fractures
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. -, no. -, 1e9, 2015 Ó Copyright 2015 Published by Elsevier Inc. on behalf of The International Society for Clinical Densitometry 1094-6950/-:1e9/$36.00 http://dx.doi.org/10.1016/j.jocd.2015.08.010
Original Article
Opportunistic Identification of Vertebral Fractures Judith E. Adams* Department of Clinical Radiology & Manchester Academic Health Science Centre, The Royal Infirmary, Central Manchester University Hospitals NHS Foundation Trust & University of Manchester, Manchester, England, United Kingdom
Abstract Vertebral fractures are powerful predictors of future fracture, so, their identification is important to ensure that patients are commenced on appropriate bone protective or bone-enhancing therapy. Risk factors (e.g., low bone mineral density and increasing age) and symptoms (back pain, loss of height) may herald the presence of vertebral fractures, which are usually confirmed by performing spinal radiographs or, increasingly, using vertebral fracture assessment with dual-energy X-ray absorptiometry scanners. However, a large number (30% or more) of vertebral fractures are asymptomatic and do not come to clinical attention. There is, therefore, scope for opportunistic (fortuitous) identification of vertebral fractures from various imaging modalities (radiographs, computed tomography, magnetic resonance imaging, and radionuclide scans) performed for other clinical indications and which include the spine in the field of view, with midline sagittal reformatted images from computed tomography having the greatest potential for such opportunistic detection. Numerous studies confirm this potential for identification but consistently find underreporting of vertebral fractures. So, a valuable opportunity to improve the management of patients at increased risk of future fracture is being squandered. Educational training programs for all clinicians and constant reiteration, stressing the importance of the accurate and clear reporting of vertebral fractures (‘‘you only see what you look for’’), can improve the situation, and automated computer-aided diagnostic tools also show promise to solve the problem of this underreporting of vertebral fractures. Key Words: Computed tomography; DXA vertebral fracture assessment; Magnetic resonance imaging; Opportunistic identification; Vertebral fracture.
vertebral fracture risk by between 30% and 70% (5,6). Identification of vertebral fractures is therefore relevant to the appropriate management of patients with osteoporosis and so at risk of further low trauma insufficiency fractures, which are associated with significant reduction in quality of life, morbidity, and mortality (7). Vertebral fractures are also relevant to the calculation of the World Health Organization 10yr fracture risk assessment tool (http://www.shef.ac.uk/ FRAX/; 8). In addition, vertebral fractures may be asymptomatic in 30% or more of subjects (9), depending on the method used to define vertebral fracture, so, imaging techniques provide the opportunity to identify the presence of vertebral fractures incidentally (fortuitously, opportunistically) when images are being performed for other and various clinical indications. These imaging methods include radiographs (lateral chest, abdominal, barium studies, intravenous urography), dual-energy X-ray absorptiometry (DXA) and subsequent vertebral fracture assessment (VFA), computed
Introduction Vertebral fractures are the most common osteoporotic fractures and occur at an earlier age than other such fractures in the humerus and hip (1). They are powerful predictors of future fracture; if a vertebral fracture is present after the age of 50 yr, the patient is at 5 times the risk of a future vertebral fracture and double the risk of a hip fracture (2e4). There are now effective bone protective and boneenhancing therapies, which for quite modest increases in bone mineral density (BMD) of 4%e12% reduce future Received 04/06/15; Accepted 08/12/15. *Address correspondence to: Judith E. Adams, MBBS, FRCP, FRCR, Department of Clinical Radiology, The Royal Infirmary, Central Manchester University Hospitals NHS Foundation Trust, Manchester, England M13 9WL, UK. E-mail: Judith.adams@ manchester.ac.uk
1
2 tomography (CT; to include lateral scout views and particularly midline sagittal reformations (SR), which involve no additional scanning or ionizing radiation exposure of the patient), magnetic resonance imaging (MRI; localizer views and direct sagittal images), and radionuclide scans (RNS; both bone and positron emission tomography CT scans). There is considerable evidence in the literature that there is underreporting by radiologists generally (10,11), as well as missing such opportunistic identification of vertebral fractures from this variety of imaging techniques (12e14). This stimulated the vertebral fracture initiatives of the International Osteoporosis Foundation, initially in collaboration with the European Society of Skeletal Radiology in 2002, and subsequently updated in 2010 to include a new section on DXA VFA. An educational resource based on this initiative is available at: http://www.iofbonehealth.org/vertebral-fracture-teaching-pro gram and includes slides, which can be downloaded for those who wish to ‘‘spread the gospel’’ of the importance to patients of the accurate identification and clear reporting of the presence of vertebral fractures (15). The clear and accurate reporting of vertebral fractures is essential. The most widely used method in the assessment and grading of vertebral fractures advocated in clinical reporting is the semiquantitative (SQ) method (16) in which changes in vertebral shape are judged subjectively, rather than using objective 6-point morphometry measurements (17). In the SQ method, 4 grades are differentiated: grade 0 5 no fracture; grade 1 5 mild fracture (reduction in vertebral height 20%e25%, compared to adjacent normal vertebrae); grade 2 5 moderate fracture (reduction in height 26%e40%); and grade 3 5 severe fracture (reduction in height more than 40%) with shape defined as predominantly wedge, end plate, or crush, but these shape abnormalities may be combined. The authors also stressed the importance of ‘‘aside from morphometric features, most vertebral fractures are readily distinguished by the presence of end plate deformities and buckling of the cortices, by the lack of parallelism of end plates, and by the loss of vertical continuity of vertebral morphology’’. They advocated ‘‘the use of a combined approach incorporating both visual and morphometric methods’’ in defining vertebral fractures in drug trials in osteoporosis. Consequently, when defining vertebral fractures in clinical practice scrutiny of the vertebral end plate, as stressed by the algorithm-based qualitative method (18), in addition to describing morphometric change in shape of the vertebral body, is essential to differentiate vertebral fractures from deformities, which may be caused by developmental short vertebral height and cupid’s bow deformity, Scheuermann disease, and spondylotic modeling (19). In a radiology report, it would improve clarity if vertebrae were regarded only as ‘‘normal,’’ ‘‘deformed,’’ or ‘‘fractured,’’ and if for the latter the grading be given, as the higher the grade and the more vertebral fractures that are present the higher the risk of future fracture. If the vertebral fractures are considered to be osteoporotic in etiology, it would also be useful to add that the appearances are those of ‘‘clinical spinal osteoporosis,’’ irrespective of what the lumbar spine DXA BMD might be, as this alerts the
Adams referring clinician to consider appropriate management strategies. DXA bone densitometry should be suggested and performed if the results will influence management. Low BMD is a risk factor for prevalent and incident vertebral fractures (20), but in a considerable number of patients with vertebral fracture, BMD may not be reduced. The terms such as ‘‘collapse,’’ ‘‘loss of height,’’ and ‘‘wedging’’ should be avoided as they do not convey to the referring clinician the significance and relevance of the features, in whatever imaging technique is being reviewed for the presence of vertebral fractures.
Imaging Techniques Radiographs Spinal radiographs, and increasingly DXA VFA, are the most widely used methods for specific imaging of clinically suspected vertebral fractures. However, the spine is included in radiographs performed for other clinical reasons. Examples include abdominal radiographs (Fig. 1A), barium studies, and intravenous urograms. The images should be scrutinized for vertebral fractures and if they are present, they should be clearly reported and anterior/posterior (AP) and lateral thoracic and lumbar spinal radiographs should be advocated to confirm their presence and grade of severity (Fig. 1B). However, it is lateral chest radiographs that have been most studied for the presence of vertebral fractures and found to be underreported (10,21e23; Fig. 1C). In a study of 934 women aged 60 yr and older and in whom a lateral chest radiograph had been performed, these were reviewed for the presence of vertebral fractures (10). Moderate or severe vertebral fractures were present in 132 (14.1%) subjects. Of these, only 50% were stated to be fractures in the radiology report and 23% in the summary, 17% had the fracture noted in the medical record or discharge summary, and in only 18% was appropriate treatment prescribed. The study indicated a need for improving recognition of opportunistic vertebral fracture identification on such imaging (10). Another study examined 10,291 women who had lateral chest radiographs and in whom 142 (1.4%) had vertebral fractures reported (22). However, in only 58 (41%) did the presence of a vertebral fracture appear in the final conclusion, in only 23 (16%) was the presence of a vertebral fracture documented in the discharge summary, and only 36% of the patients were using any osteoporosis medications at discharge. The authors concluded that vertebral fractures from lateral chest radiographs represented a missed opportunity for osteoporosis management (22). In a smaller number of women (106) of various ethnicities with a mean age of 65 (range 55e89) yr, the lateral chest radiographs were reviewed with a 1e2-yr follow-up (23). Twenty-six of 106 patients (25%) had vertebral fractures; the fracture prevalence increased with age (in 17 of 54 [13%] women under the age of 65 yr; in 19 of 52 [37%] women older than 65 yr), and 3 of 16 (19%) developed interval fractures. In only 4 of 26 (15%) patients was the fracture included in the report; and although 31 of 106 (29%) were scheduled for bone densitometry, this was performed in only 6 of 106 (6%)
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures
3
Fig. 1. Radiographs: vertebral fractures can opportunistically be identified from any radiographs performed for various clinical indications and which include the spine in the field of view. (A) Anteroposterior (AP) abdominal radiograph showing vertebral fractures of T11 and T12, which are reduced in height and have increased density of the end plates. In such cases, dedicated AP and lateral thoracic and lumbar spinal radiographs should be performed. (B) Lateral lumbar spinal radiograph confirms grade 3 severe vertebral fractures of T11 and T12 and shows additional vertebral fractures of L2, L3, and L4. (C) Lateral chest radiograph with a grade 3 severe wedge fracture of T6 and additional fractures of T11 and T2; the appearances indicate clinical spinal osteoporosis. patients (23). The lateral chest radiograph, particularly in an elderly population, therefore offers an opportunity to identify vertebral fractures, which may be present and which will impact on patient management. However, this will require
the radiologist to actively scrutinize the image for vertebral fractures being present and then report them clearly as fractures. There is evidence that simple training programs improve such vertebral fracture identification (24,25).
Fig. 2. Dual-energy X-ray absorptiometry: (A) Lumbar spine posterior-anterior (PA) bone mineral density scans that show vertebral fractures of T12 and L2, which are reduced in height with sclerosis and concavity of the upper end plates. With such findings, anteroposterior and (B) lateral spinal radiographs should be performed to confirm the presence of fractures; there is a grade 3 severe crush fracture of T12 and a grade 1 mild upper end plate of L2. Alternatively, if appropriate software is available, then (C) DXA lateral vertebral fracture assessment can be performed. In this different patient, there is a grade 3 severe crush fracture of T12. Note aortic calcification anterior to the lumbar spine from which aortic calcification scoring can be derived and which is related to cardiac morbidity and mortality. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4
Adams
Dual-Energy X-Ray Absorptiometry
Computed Tomography
Vertebral fractures can be evident on the posterior-anterior (PA) DXA scan performed to measure BMD (Fig. 2A). There may be reduction in vertical height and area of the vertebral body affected, and consequently higher BMD, particularly when compared with any previous DXA examinations. With the technical developments in DXA (improved sensitivity of detectors; small increase in X-ray exposure and increased spatial resolution [0.50e0.35 mm]), image quality has improved. This enables better visualization of the vertebral end plate changes, which occur with vertebral fractures. If vertebral fractures are present on the lumbar spine PA BMD scan, then either anteroposterior (AP) and lateral spinal radiographs (Fig. 2B) or DXA VFA (Fig. 2C), if relevant software is available on the scanner, should be performed to confirm the presence of vertebral fracture as this will influence patient management. Alternatively, a protocol could be put in place to perform VFA at the same time as DXA for BMD is performed, which makes scheduling of appointments easier. In our unit, we perform VFA in patients referred for BMD in women aged more than 65 yr, men aged more than 70 yr and all who have had significant (O5 mg/d for longer than 3 mo) oral glucocorticoid therapy over 45 yr, and who have not had any spinal imaging in the previous past 12 mo. Such a protocol can enhance vertebral fracture identification and so improve targeted treatment and management (26).
CT scans are performed widely for a variety of clinical indications. Preceding each CT examination preliminary scan projection radiographs (scout views), generally AP and lateral, are performed to define the anatomical area to be scanned with transverse axial images. The lateral scout image should be carefully scrutinized for vertebral fractures (Fig. 3A). In the Framingham Osteoporosis Study, participants included 50 women and 50 men (age 50e87 yr, mean 70 yr) (27). On lateral CT scout views, T4eL4 vertebrae were assessed independently by 2 radiologists on 2 occasions using an SQ scale as normal, mild, moderate, or severe fracture. The vertebraspecific prevalence of grade 1 (mild) fracture ranged from 3% to 5%.The inter-reader agreement for grade 1 vertebral fractures was fair (k 5 56%e59%) and good (k 5 68% e72%) for grade 2 fractures. Intra-reader agreement for grade 1 vertebral fracture was fair (k 5 55%) for 1 reader and excellent for another (k 5 77%), and intra-reader agreement for grade 2 vertebral fracture was excellent for both readers (k 5 76% and 98%). Thoracic vertebrae were more difficult to evaluate than those in the lumbar region, and agreement was lowest (inter-reader k 5 43%) for fracture at the upper (T4eT9) thoracic levels and highest (inter-reader k 5 76% e78%) for the lumbar spine (L1eL4). The authors concluded that lateral CT scout views can be useful to assess for vertebral fracture (27). In another study of 500 patients (303 males, 197
Fig. 3. Computed tomography: (A) Lateral scout view demonstrating grade 2 moderate upper end plate fracture of L1. (B) Midline sagittal reformation (SR), which can be derived from all MDCT scans with no additional scanning having to be performed or ionizing exposures to the patient, showing grade 3 severe fractures of T3 and T7 in a different patient. Such images are particularly sensitive to identifying vertebral fractures because the middle of the end plate is weaker than the peripheral ring, and will demonstrate fractures not evident on the transverse axial images, so it is essential that such SR images are routinely acquired and always carefully scrutinized for fractures. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures females, aged 64.6 13.5 yr old) who had CT scans, the identification of vertebral fractures from the CT scout view was compared with CT midline SR (28). In a significant number of scans (59 of 500 [11.8%]), the evaluation on scout CT was incomplete or limited for patient/technical-based conditions (obesity, scoliosis). In 67 of 485 patients (13.8%), 99 vertebral fractures were detected of which only 18 (26.9%) were included in the report. Scrutiny of the scout view took only 41 s and is advocated by the authors to identify vertebral fractures (28). Members of the same group reviewed 300 CT examinations of the thoracic and/or lumbar spine, which were collected and independently analyzed by 3 musculoskeletal radiologists in 2 different sessions (29). A total of 1522 vertebrae were considered (130 males and 170 females; ages, 73.0 2.8 yr). In 73 of 1522 (4.8%), vertebral fractures were identified in 34 of 300 patients (11.3%). The sensitivity and specificity of CT scout views for identification of vertebral fractures were 98.7% and 99.7%, respectively. Accuracy (Area under the Receiver Operator Characteristics [AUROC] 5 0.992 0.008) and interobserver agreement (k 5 0.968 0.008) were excellent. Intraobserver agreement was perfect (k 5 1.000). Performance of this method was independent of spondylosis, vertebral level, and type and grade of vertebral fractures. The authors concluded that CT scout views are a simple but very accurate method for the detection of vertebral fractures (29). With the technical developments in CT (spiral rotation of the X-ray tube; multiple detectors; multidetector spiral CT; MDCT), which have occurred over the past decade, scanning is extremely fast (the torso is imaged in less than 20 s), and spatial resolution improved (0.6 mm) with 3D volumetric or 2D coronal/SR acquired rapidly. These reformatted images require no additional scanning to be performed and involve no additional exposure of the patient to ionizing radiation. The peripheral ring of the vertebral body end plate is the strongest part and the central area the weakest, and it is in this central area where insufficiency microfractures may occur first. This makes midline sagittal CT reformations the most sensitive method for identifying vertebral fractures (Fig. 3B), and which will not be evident on the transverse axial sections. MDCT of the thorax and abdomen is now performed widely worldwide for various clinical indications and is established as probably the most useful method to opportunistically identify vertebral fractures, but again, there is evidence that there is underreporting of these fractures. In an early study of 200 consecutive MDCT with midline SR in patients aged 61 (range 18e92) yr (48% women), there were 70 (35%) with vertebral fractures, of which 51 (73%) were mild grade 1, 13 (19%) were moderate grade 2, and 6 (9%) were severe grade 3 fractures. However, only 6 (9%) were included in the radiology report (30). In a similar study, consecutive MDCT scans performed in 192 patients (95 women) with a mean age of 70.1 yr using slice thickness of 1.25 mm (13) were reviewed. There were 38 (19.8%) patients who had 1 or more grade 2 moderate or grade 3 severe vertebral fractures, but only 5 (11%) were recorded as osteoporotic vertebral fractures in the formal report. The sensitivity of
5 identification of the vertebral fracture on the axial sections was low at 0.35 (13). In another study, abdominal or thoracoabdominal MDCT was performed in 112 postmenopausal women, with axial images and SR being analyzed separately by 2 radiologists. In 27 patients, osteoporotic vertebral fractures were found; only 6 of these could be identified on the axial images, but none of these were included in the radiology report. The authors concluded that SR of standard MDCT images provide important additional information on spinal abnormalities; in particular, osteoporotic vertebral fractures are substantially better detected (31). In a further study, 323 consecutive patients (196 males, 127 females) with a mean age of 62.6 yr (range 20e88) who had chest and/or abdominal MDCT were evaluated (14). Sagittal reformats of the spine obtained from thin section data sets were reviewed by 2 radiologists and assessed for vertebral fractures. A vertebral body height loss of 15% or more was considered a fracture and graded as mild (15%e24%), moderate (25%e49%), or severe (more than 50%). Thirty-one of 323 patients (9.5%) had at least 1 vertebral fracture, and 7 of those patients had multiple fractures with a total of 41 fractures. Morphometric grading revealed 10 mild, 16 moderate, and 15 severe fractures. Prevalence was higher in women (14.1%) than men (6.6%) and increased with patient’s age with 17.1% prevalence in postmenopausal women. Only 6 of 41 vertebral fractures (14.6%) had been noted in the radiology final report, whereas the remaining 35 (85.4%) had not (14). In an audit study, in 175 consecutive patients aged O65 yr, sagittal reformatted images of the spine were obtained from CT scans of the chest or abdomen (32). Vertebral fractures were defined using an SQ technique. The prevalence of vertebral fractures was 13%, with 41 vertebral fractures identified in 22 patients; 12 of 22 (55%) had vertebral fracture included in the formal CT report. The vertebral fracture was newly identified in 17 (77%) patients, but vertebral fracture and osteoporosis were each listed in the relevant discharge summary or clinic letter for only 14% of patients, and only 31% of patients with fracture subsequently received osteoporosis treatment. The authors concluded that sagittal reformatted images of the spine from CT scans of the chest or abdomen detect previously unidentified vertebral fractures and offer an undervalued opportunity to assess fracture risk and intervene with treatments that prevent fractures and reduce mortality. In an in vitro study using 65 vertebrae from 21 human cadavers spines using 64-row multidetector CT scanner, axial images were acquired with slice thicknesses of 0.6, 1, 2, 3, and 5 mm, and SR were obtained using these data sets (33). The specimens were also radiographed in AP and lateral orientation. Vertebrae visualized in the different image data sets were separately graded by 4 radiologists. Fracture status determined in a consensus reading of interactive reformations of the 0.6-mm CT data set in all 3 dimensions served as a standard of reference in combination with pathological examinations. The average agreement for the 0.6-mm SR obtained between each radiologist and standard of reference for the grading of the fractures was very good (k 5 0.81). It was good for the 1-, 2- and 3-mm SR (k 5 0.70, 0.69 and
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6
Adams
0.64), but only moderate for the radiographs (k 5 0.52), and fair for the 5-mm SR (k 5 0.33). When considering only the detection of fractures, independent of the grading, all kappa values improved by about 0.15, resulting in excellent values for the 0.6 mm through 3-mm SR (0.95 ! k ! 0.79) and good values for the radiographs (k 5 0.72). Ninety-five percent of the fractures could be identified using the 1-mm SR, but 18% of the fractures were missed on the radiographs. The authors concluded that sagittal CT reformations could more accurately assess vertebral fractures than standard spinal radiographs. However, for reliable detection of these fractures, SR derived from axial images with a slice thickness of 3 mm or less are required. The thinnest available axial slice thickness (0.6 mm) performed best in fracture grading, and this is what is generally used in clinical practice (33).
Magnetic Resonance Imaging MRI plays an important role in differentiating acute from long-standing vertebral fractures by revealing the presence of marrow edema in acute fractures and restoration of marrow fat signal in old. MRI is also the imaging method of choice to differentiate vertebral fracture due to benign and malignant etiologies, with the latter having altered soft-tissue signal in the vertebral body, convexity of the posterior vertebral margin, cortical destruction, a soft-tissue mass, and high signal on diffusion-weighted sequences (19). Unlike CT midline sagittal magnetic resonance (MR) images using specific sequences are acquired directly and do not need to be reformatted. Vertebral fractures may be identified opportunistically on MRI scans, which have been performed for various and other clinical indications and include the spine. Before an MR scan, a localizer scan is performed to define the anatomical location of the other scan sequences to be performed. Localizer scans are derived from thick slices in 3 planes (axial, coronal, and sagittal) and so are quite rudimentary images that are limited in spatial resolution and image quality. Despite this poor image quality, there is evidence that such localizer images can be used to identify vertebral fracture. However, subsequently acquired directly sagittal images using specifically selected sequences (Fig. 4) give much higher quality images than the localizer. In a study of 300 MR scans (147 males, 153 females, 59.4 16.4 yr) of the thoracic and/ or lumbar spine, a total of 2186 vertebrae were assessed (34). The initial assessment was by an SQ approach, and morphometric analysis was performed when a vertebral fracture was suspected on the MR localizer images evaluated by 3 radiologists on 2 separate occasions. A full diagnostic sagittal T1weighted fast spin echo MR sequence was used as standard of reference for identification of vertebral fractures. Sixty-seven of 2136 (3.1%) vertebral fractures were identified in 23 of 300 (7.7%) patients. In detection of fractures, sensitivity and specificity of the MR localizer were both 100% (accuracy AUROC 5 1.000). Interobserver agreement was excellent (k 5 0.938 0.013), whereas intraobserver agreement was perfect (k 5 1.000). The diagnostic performance was independent of vertebral level, type, and grade of fractures. In another study, MR localizer sagittal images of 856 patients
Fig. 4. Magnetic resonance imaging: although vertebral fractures can be identified on the initial localizer views, these are of limited image quality. Subsequently acquired direct sagittal images using specifically selected sequences give much higher quality images than the localizer. T2-weighted sagittal image confirms grade 3 severe wedge fractures of T7 and T9; the latter shows high signal in the marrow indicating edema and that the fracture is recent. The fracture at T9 has regained normal marrow fat signal indicating that it is an old fracture. undergoing breast MRI were reviewed by 3 expert musculoskeletal radiologists with an SQ approach to detecting vertebral fractures (35). Fifty-seven MR localizer images (57 of 856; 6.7%) were inadequate for diagnostic purposes. MR localizer images detected vertebral fractures in 71 of 799 patients (8.9%); of these, 63 of 71 (88.7%) vertebral fractures were not reported. The authors concluded that MR localizer images should always be scrutinized for the presence of vertebral fractures in an MRI scan being performed for other clinical indications.
Radionuclide Imaging RNS with technetium (99mTc) bisphosphonates is a good skeletal survey technique to identify areas of abnormal uptake of isotope, which generally reflects increased blood flow, bone turnover, and bone formation. The technique is very sensitive, but not specific in that numerous pathologies (osteoarthritis, osteomyelitis, blastic bone metastases, Paget’s disease of bone, fractures, Looser zones) result in increased uptake.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures Acute vertebral fractures will have increased uptake, which is usually apparent within 48 h of fracture and will persist for between 6 and 18 mo (36,37; Fig. 5). A fracture may therefore be an incidental finding in a patient undergoing RNS for other clinical indications, for example, for bone metastases in known primary cancer. However, having cancer does not preclude the patient having insufficiency fractures and/ or osteoporosis. Positron emission tomography CT with F18 fluorodeoxyglucose is increasingly used in patients with a variety of cancers for tumor staging. There may be increased uptake of radionuclide in osteoporotic fractures, which must therefore be differentiated from bone metastases (38).
Discussion From the peer-reviewed literature, there is good evidence that many imaging techniques being performed for various clinical indications offer the opportunity to identify vertebral fractures incidentally if the spine is in the field of view. These include radiographs (10e12,21e23), CT scout views and SR (13,14,27e33), MRI localizers, and sagittal images using specific sequences (19,34,35) and radionuclide bone scans (36e38). Probably the most sensitive is midline sagittal CT reformations, which are very rapidly (a few seconds) acquired by technical staff at the time of the scan, and this should be routine practice in radiology departments. This is more efficient and preferable to leaving the choice to obtain and perform such reformats to the radiologist at the time of reporting. Such midline SR will identify vertebral fractures, which are not evident on transverse axial images (13,31), so, it is essential for these reformatted images to be viewed and scrutinized at the time of reporting, otherwise vertebral fractures will be missed. MDCT is now very widely performed, especially in elderly patients in whom the prevalence of vertebral
Fig. 5. Radionuclide bone scan: such a scan provides a good survey technique of the whole skeleton is sensitive but not very specific and is performed for a variety of clinical indications. PA image: there is increased uptake in the body of T7 due to an acute vertebral fracture. Such fractures show increased uptake within 24 h of the fracture occurring, but the increased uptake may persist for up to 2 yr.
7 fractures will be higher than in a younger cohort. However, although vertebral fractures are present on an image, there is evidence that there is widespread and worldwide underreporting in all imaging techniques (10e15). In a review of the literature in 2010 from 12 studies (7 using radiographs; 5 using MDCT), reporting rates were low with a mean value of 27.4% (range 0%e66.3%) and were significantly lower in MDCT than in radiographs (mean 8.1% vs 41.2%; 12). So, although scientific studies have shown a high prevalence of opportunistically imaged vertebral fractures, the very significant underreporting of such fractures will have a detrimental effect on patient management. This underreporting occurs despite the vertebral fracture initiative of the International Osteoporosis Foundation having been launched over a decade ago. The problem is largely that ‘‘observers will only see what they look for’’ and spine images are not being scrutinized by radiologists. This is likely to be due to them reporting CT scans in their particular area of special expertise (gastrointestinal, cardiothoracic, and so forth) and so being unaware of the importance and relevance of vertebral fractures and their significance to the appropriate management of patients with osteoporosis. Constant and reiterative teaching and training of radiologists to emphasize the importance of reporting vertebral fractures is required (24,25). Another potential solution might be the use of automated computeraided diagnostic tools to prompt the radiologist to recognize that vertebral fractures may be present in the image being reviewed. A commercial tool is available (SpineAnalyzer, Optasia Medical, Cheadle, UK), which provides a semiautomated quantitative vertebral morphometry devised from shape-based statistical modeling and has been applied in several studies (39e41). In 1 report, based on the Framingham Heart Study Offspring and Third Generation MultiDetector CT Study (39), 1246 vertebrae were analyzed on lateral CT scout views in 96 subjects, and the authors concluded that the semiautomated algorithm provided excellent intra- and inter-reader reliability for vertebral height measurements, along with good to fair reliability for vertebral height ratios. Reliability for VFA based solely on quantitative morphometry was also good and was comparable to previous reports for SQ vertebral fracture grading by radiologists (39). In a report from the population-based Rotterdam study, which includes 14,926 inhabitants aged 45 yr and older in a subset of 99 lateral spinal radiographs, Cobb’s kyphosis angle was calculated manually and using SpineAnalyzer. The authors conclude that such vertebral fracture morphometric data to derive the Cobb’s kyphosis angle are relatively reliable, and that additional data (vertebral wedging, intervertebral disc space, spondylolisthesis, and the lordosis angle) can be derived. The average time to complete the semiautomated morphometry analysis was approximately 9 min 40 s less than previously reported for manual morphometry analysis (40), and so may be particularly useful in application to large osteoporosis epidemiological or therapeutic efficacy studies. SpineAnalyzer has also been applied in the Mr Os study (41), a cohort of community-dwelling men aged 65 yr in whom lateral spinal radiographs were available in 5958 of
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
8
Adams
5994 participants at visit 1, and 4399 of 4423 participants at Visit 2 after 4.6 yr. A subset of 494 subjects’ images was read manually to assess the sensitivity and specificity of triage by the semiautomated tool. Triage identified 3215 (53.9%) participants with radiographs that required further evaluation by the physician reader. For prevalent fractures at visit 1 (SQ 1), intra-reader kappa statistics ranged from 0.79 to 0.92; percent agreement ranged from 96.9% to 98.9%; sensitivity of the triage was 96.8%, and specificity of triage was 46.3%. The authors concluded that this triage process reduced expert reader workload without hindering the ability to identify vertebral fractures (41). Other groups have reported semiautomated tools for vertebral fracture identification in MDCT (42), in DXA VFA images (43), and spinal radiographs (44), which show potential. With further developments in computer vision techniques (45), it may become feasible to implement a completely automated search of a radiology picture archiving and communication systems (PACS) for images that contain vertebral fracture, which would be ideal. There is evidence that the presence of incidentally identified and reported vertebral fractures improves rates of bone densitometry performed and treatments implemented by clinicians (46), and that such a strategy is cost-effective (47). In conclusion there is a great potential to identify vertebral fractures opportunistically using a variety of imaging techniques, which include the spine in the field of view and which are being performed for other and varied clinical reasons. This is beneficial to patient management, is cost effective, and so has the potential to reduce future fracture risk. Unfortunately, however, there are numerous reports of the underreporting of such fractures by radiologists. Educational and training initiatives are beneficial to all clinicians, but constant reiteration is required to reinforce the message of the importance and relevance of accurate and clear reporting of vertebral fractures, particularly by radiologists, to determine appropriate patient management. To this end, it is essential that radiologists always scrutinize any images of the spine for vertebral fracture, as what is not sought will not be evident. Semiautomated/automated computer tools show potential to improve the identification of vertebral fractures and reduce the amount of manual interpretation required in large epidemiological and therapeutic efficacy trials by using such tools in a triage protocol.
Acknowledgments The author has nothing to disclose.
References 1. Christiansen BA, Bouxsein ML. 2010 Biomechanics of vertebral fractures and the vertebral fracture cascade. Curr Osteoporos Rep 8(4):198e204. 2. Delmas PD, Genant HK, Crans GG, et al. 2003 Severity of prevalent vertebral fractures and the risk of subsequent vertebral and nonvertebral fractures: results from the MORE trial. Bone 33(4): 522e532.
3. Klotzbuecher CM, Ross PD, Landsman PB, et al. 2000 Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 15(4):721e739. 4. Lindsay R, Silverman SL, Cooper C, et al. 2001 Risk of new vertebral fracture in the year following a fracture. JAMA 285(3):320e323. 5. Crandall CJ, Newberry SJ, Diamant A, et al. 2014 Comparative effectiveness of pharmacologic treatments to prevent fractures: an updated systematic review. Ann Intern Med 161(10): 711e723. 6. Reid IR. 2015 Efficacy, effectiveness and side effects of medications used to prevent fractures. J Intern Med 277(6):690e706. 7. Johnell O, Kanis JA. 2006 An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 17(12):1726e1733. 8. Kanis JA, Oden A, Johansson H, et al. 2009 FRAX and its applications to clinical practice. Bone 44(5):734e743. 9. Ensrud KE. 2013 Epidemiology of fracture risk with advancing age. J Gerontol A Biol Sci Med Sci 68(10):1236e1242. 10. Gehlbach SH, Bigelow C, Heimisdottir M, et al. 2000 Recognition of vertebral fracture in a clinical setting. Osteoporos Int 11(7):577e582. 11. Delmas PD, van de Langerijt L, Watts NB, et al. 2005 Underdiagnosis of vertebral fractures is a worldwide problem: the IMPACT study. J Bone Miner Res 20(4):557e563. 12. Bartalena T, Rinaldi MF, Modolon C, et al. 2010 Incidental vertebral compression fractures in imaging studies: lessons not learned by radiologists. World J Radiol 2(10):399e404. 13. Williams AL, Al-Busaidi A, Sparrow PJ, et al. 2009 Underreporting of osteoporotic vertebral fractures on computed tomography. Eur J Radiol 69(1):179e183. 14. Bartalena T, Giannelli G, Rinaldi MF, et al. 2009 Prevalence of thoracolumbar vertebral fractures on multidetector CT: underreporting by radiologists. Eur J Radiol 69(3):555e559. 15. Lentle BC, Brown JP, Khan A, et al. 2007 Recognizing and reporting vertebral fractures: reducing the risk of future osteoporotic fractures. Can Assoc Radiol J 58(1):27e36. 16. Genant HK, Wu CY, van Kuijk C, Nevitt MC. 1993 Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 8(9):1137e1148. 17. Guglielmi G, Diacinti D, van Kuijk C, et al. 2008 Vertebral morphometry: current methods and recent advances. Eur Radiol 18(7):1484e1496. 18. Jiang G, Eastell R, Barrington NA, Ferrar L. 2004 Comparison of methods for the visual identification of prevalent vertebral fracture in osteoporosis. Osteoporos Int 15(11):887e896. 19. Link TM, Guglielmi G, van Kuijk C, Adams JE. 2005 Radiologic assessment of osteoporotic vertebral fractures: diagnostic and prognostic implications. Eur Radiol 15(8):1521e1532. 20. Cauley JA, Zmuda JM, Wisniewski SR, et al. 2004 Bone mineral density and prevalent vertebral fractures in men and women. Osteoporos Int 15(1):32e37. 21. Kim N, Rowe BH, Raymond G, et al. 2004 Underreporting of vertebral fractures on routine chest radiography. AJR Am J Roentgenol 182(2):297e300. J Gen Intern Med. 2006 Apr; 21(4):352-356. 22. Morris CA, Carrino JA, Lang P, Solomon DH. 2006 Incidental vertebral fractures on chest radiographs. Recognition, documentation, and treatment. J Gen Intern Med 21(4):352e356. 23. Mui LW, Haramati LB, Alterman DD, et al. 2003 Evaluation of vertebral fractures on lateral chest radiographs of inner-city postmenopausal women. Calcif Tissue Int 73(6):550e554. 24. Casez P, Uebelhart B, Gaspoz JM, et al. 2006 Targeted education improves the very low recognition of vertebral fractures and
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures
25. 26.
27.
28. 29. 30. 31.
32. 33. 34. 35.
osteoporosis management by general internists. Osteoporos Int 17(7):965e970. Gruber M, Dinges J, M€uller D, et al. 2013 Impact of specific training in detecting osteoporotic vertebral fractures on routine chest radiographs. Rofo 185(11):1074e1080. Kuet KP, Charlesworth D, Peel NF. 2013 Vertebral fracture assessment scans enhance targeting of investigations and treatment within a fracture risk assessment pathway. Osteoporos Int 24(3):1007e1014. Samelson EJ, Christiansen BA, Demissie S, et al. 2011 Reliability of vertebral fracture assessment using multidetector CT lateral scout views: the Framingham Osteoporosis Study. Osteoporos Int 22(4):1123e1131. Bazzocchi A, Spinnato P, Albisinni U, et al. 2012 A careful evaluation of scout CT lateral radiograph may prevent unreported vertebral fractures. Eur J Radiol 81(9):2353e2357. Bazzocchi A, Fuzzi F, Garzillo G, et al. 2013 Reliability and accuracy of scout CT in the detection of vertebral fractures. Br J Radiol 86(1032):20130373. Woo EK, Mansoubi H, Alyas F. 2008 Incidental vertebral fractures on multidetector CT images of the chest: prevalence and recognition. Clin Radiol 63(2):160e164. M€uller D, Bauer JS, Zeile M, et al. 2008 Significance of sagittal reformations in routine thoracic and abdominal multislice CT studies for detecting osteoporotic fractures and other spine abnormalities. Eur Radiol 18(8):1696e1702. Chan PL, Reddy T, Milne D, Bolland MJ. 2012 Incidental vertebral fractures on computed tomography. N Z Med J 125(1350): 45e50. Bauer JS, M€uller D, Ambekar A, et al. 2006 Detection of osteoporotic vertebral fractures using multidetector CT. Osteoporos Int 17(4):608e615. Bazzocchi A, Garzillo G, Fuzzi F, et al. 2014 Localizer sequences of magnetic resonance imaging accurately identify osteoporotic vertebral fractures. Bone 61:158e163. Bazzocchi A, Spinnato P, Garzillo G, et al. 2012 Detection of incidental vertebral fractures in breast imaging: the potential role of MR localisers. Eur Radiol 22(12):2617e2623.
9 36. Matin P. 1979 The appearance of bone scans following fractures, including immediate and long-term studies. J Nucl Med 20(12): 1227e1231. 37. Ryan PJ, Fogelman I. 1994 Osteoporotic vertebral fractures: diagnosis with radiography and bone scintigraphy. Radiology 190(3):669e672. 38. Katzel JA, Heiba SI. 2005 PET/CT F-18 FDG scan accurately identifies osteoporotic fractures in a patient with known metastatic colorectal cancer. Clin Nucl Med 30(10):651e654. 39. Kim YM, Demissie S, Eisenberg R, et al. 2011 Intra-and inter-reader reliability of semi-automated quantitative morphometry measurements and vertebral fracture assessment using lateral scout views from computed tomography. Osteoporos Int 22(10):2677e2688. 40. Oei L, Ly F, El Saddy S, et al. 2013 Multi-functionality of computer-aided quantitative vertebral fracture morphometry analyses. Quant Imaging Med Surg 3(5):249e255. 41. Cawthon PM, Haslam J, Fullman R, et al. 2014 Methods and reliability of radiographic vertebral fracture detection in older men: the osteoporotic fractures in men study. Bone 67:152e155. 42. Baum T, Bauer JS, Klinder T, et al. 2014 Automatic detection of osteoporotic vertebral fractures in routine thoracic and abdominal MDCT. Eur Radiol 24(4):872e880. 43. Roberts MG, Pacheco EM, Mohankumar R, et al. 2010 Detection of vertebral fractures in DXA VFA images using statistical models of appearance and a semi-automatic segmentation. Osteoporos Int 21(12):2037e2046. 44. Roberts MG, Oh T, Pacheco EM, et al. 2012 Semi-automatic determination of detailed vertebral shape from lumbar radiographs using active appearance models. Osteoporos Int 23(2):655e664. 45. Roberts MG, Cootes TF, Adams JE. 2012 Automatic location of vertebrae on DXA images using random forest regression. Med Image Comput Comput Assist Interv 15(Pt 3):361e368. 46. Majumdar SR, McAlister FA, Johnson JA, et al. 2012 Interventions to increase osteoporosis treatment in patients with ‘incidentally’ detected vertebral fractures. Am J Med 125(9):929e936. 47. Majumdar SR, Lier DA, McAlister FA, et al. 2013 Costeffectiveness of osteoporosis interventions for ‘incidental’; vertebral fractures. Am J Med 126(2):169.e9e169.e17.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Original Article
Opportunistic Identification of Vertebral Fractures Judith E. Adams* Department of Clinical Radiology & Manchester Academic Health Science Centre, The Royal Infirmary, Central Manchester University Hospitals NHS Foundation Trust & University of Manchester, Manchester, England, United Kingdom
Abstract Vertebral fractures are powerful predictors of future fracture, so, their identification is important to ensure that patients are commenced on appropriate bone protective or bone-enhancing therapy. Risk factors (e.g., low bone mineral density and increasing age) and symptoms (back pain, loss of height) may herald the presence of vertebral fractures, which are usually confirmed by performing spinal radiographs or, increasingly, using vertebral fracture assessment with dual-energy X-ray absorptiometry scanners. However, a large number (30% or more) of vertebral fractures are asymptomatic and do not come to clinical attention. There is, therefore, scope for opportunistic (fortuitous) identification of vertebral fractures from various imaging modalities (radiographs, computed tomography, magnetic resonance imaging, and radionuclide scans) performed for other clinical indications and which include the spine in the field of view, with midline sagittal reformatted images from computed tomography having the greatest potential for such opportunistic detection. Numerous studies confirm this potential for identification but consistently find underreporting of vertebral fractures. So, a valuable opportunity to improve the management of patients at increased risk of future fracture is being squandered. Educational training programs for all clinicians and constant reiteration, stressing the importance of the accurate and clear reporting of vertebral fractures (‘‘you only see what you look for’’), can improve the situation, and automated computer-aided diagnostic tools also show promise to solve the problem of this underreporting of vertebral fractures. Key Words: Computed tomography; DXA vertebral fracture assessment; Magnetic resonance imaging; Opportunistic identification; Vertebral fracture.
vertebral fracture risk by between 30% and 70% (5,6). Identification of vertebral fractures is therefore relevant to the appropriate management of patients with osteoporosis and so at risk of further low trauma insufficiency fractures, which are associated with significant reduction in quality of life, morbidity, and mortality (7). Vertebral fractures are also relevant to the calculation of the World Health Organization 10yr fracture risk assessment tool (http://www.shef.ac.uk/ FRAX/; 8). In addition, vertebral fractures may be asymptomatic in 30% or more of subjects (9), depending on the method used to define vertebral fracture, so, imaging techniques provide the opportunity to identify the presence of vertebral fractures incidentally (fortuitously, opportunistically) when images are being performed for other and various clinical indications. These imaging methods include radiographs (lateral chest, abdominal, barium studies, intravenous urography), dual-energy X-ray absorptiometry (DXA) and subsequent vertebral fracture assessment (VFA), computed
Introduction Vertebral fractures are the most common osteoporotic fractures and occur at an earlier age than other such fractures in the humerus and hip (1). They are powerful predictors of future fracture; if a vertebral fracture is present after the age of 50 yr, the patient is at 5 times the risk of a future vertebral fracture and double the risk of a hip fracture (2e4). There are now effective bone protective and boneenhancing therapies, which for quite modest increases in bone mineral density (BMD) of 4%e12% reduce future Received 04/06/15; Accepted 08/12/15. *Address correspondence to: Judith E. Adams, MBBS, FRCP, FRCR, Department of Clinical Radiology, The Royal Infirmary, Central Manchester University Hospitals NHS Foundation Trust, Manchester, England M13 9WL, UK. E-mail: Judith.adams@ manchester.ac.uk
1
2 tomography (CT; to include lateral scout views and particularly midline sagittal reformations (SR), which involve no additional scanning or ionizing radiation exposure of the patient), magnetic resonance imaging (MRI; localizer views and direct sagittal images), and radionuclide scans (RNS; both bone and positron emission tomography CT scans). There is considerable evidence in the literature that there is underreporting by radiologists generally (10,11), as well as missing such opportunistic identification of vertebral fractures from this variety of imaging techniques (12e14). This stimulated the vertebral fracture initiatives of the International Osteoporosis Foundation, initially in collaboration with the European Society of Skeletal Radiology in 2002, and subsequently updated in 2010 to include a new section on DXA VFA. An educational resource based on this initiative is available at: http://www.iofbonehealth.org/vertebral-fracture-teaching-pro gram and includes slides, which can be downloaded for those who wish to ‘‘spread the gospel’’ of the importance to patients of the accurate identification and clear reporting of the presence of vertebral fractures (15). The clear and accurate reporting of vertebral fractures is essential. The most widely used method in the assessment and grading of vertebral fractures advocated in clinical reporting is the semiquantitative (SQ) method (16) in which changes in vertebral shape are judged subjectively, rather than using objective 6-point morphometry measurements (17). In the SQ method, 4 grades are differentiated: grade 0 5 no fracture; grade 1 5 mild fracture (reduction in vertebral height 20%e25%, compared to adjacent normal vertebrae); grade 2 5 moderate fracture (reduction in height 26%e40%); and grade 3 5 severe fracture (reduction in height more than 40%) with shape defined as predominantly wedge, end plate, or crush, but these shape abnormalities may be combined. The authors also stressed the importance of ‘‘aside from morphometric features, most vertebral fractures are readily distinguished by the presence of end plate deformities and buckling of the cortices, by the lack of parallelism of end plates, and by the loss of vertical continuity of vertebral morphology’’. They advocated ‘‘the use of a combined approach incorporating both visual and morphometric methods’’ in defining vertebral fractures in drug trials in osteoporosis. Consequently, when defining vertebral fractures in clinical practice scrutiny of the vertebral end plate, as stressed by the algorithm-based qualitative method (18), in addition to describing morphometric change in shape of the vertebral body, is essential to differentiate vertebral fractures from deformities, which may be caused by developmental short vertebral height and cupid’s bow deformity, Scheuermann disease, and spondylotic modeling (19). In a radiology report, it would improve clarity if vertebrae were regarded only as ‘‘normal,’’ ‘‘deformed,’’ or ‘‘fractured,’’ and if for the latter the grading be given, as the higher the grade and the more vertebral fractures that are present the higher the risk of future fracture. If the vertebral fractures are considered to be osteoporotic in etiology, it would also be useful to add that the appearances are those of ‘‘clinical spinal osteoporosis,’’ irrespective of what the lumbar spine DXA BMD might be, as this alerts the
Adams referring clinician to consider appropriate management strategies. DXA bone densitometry should be suggested and performed if the results will influence management. Low BMD is a risk factor for prevalent and incident vertebral fractures (20), but in a considerable number of patients with vertebral fracture, BMD may not be reduced. The terms such as ‘‘collapse,’’ ‘‘loss of height,’’ and ‘‘wedging’’ should be avoided as they do not convey to the referring clinician the significance and relevance of the features, in whatever imaging technique is being reviewed for the presence of vertebral fractures.
Imaging Techniques Radiographs Spinal radiographs, and increasingly DXA VFA, are the most widely used methods for specific imaging of clinically suspected vertebral fractures. However, the spine is included in radiographs performed for other clinical reasons. Examples include abdominal radiographs (Fig. 1A), barium studies, and intravenous urograms. The images should be scrutinized for vertebral fractures and if they are present, they should be clearly reported and anterior/posterior (AP) and lateral thoracic and lumbar spinal radiographs should be advocated to confirm their presence and grade of severity (Fig. 1B). However, it is lateral chest radiographs that have been most studied for the presence of vertebral fractures and found to be underreported (10,21e23; Fig. 1C). In a study of 934 women aged 60 yr and older and in whom a lateral chest radiograph had been performed, these were reviewed for the presence of vertebral fractures (10). Moderate or severe vertebral fractures were present in 132 (14.1%) subjects. Of these, only 50% were stated to be fractures in the radiology report and 23% in the summary, 17% had the fracture noted in the medical record or discharge summary, and in only 18% was appropriate treatment prescribed. The study indicated a need for improving recognition of opportunistic vertebral fracture identification on such imaging (10). Another study examined 10,291 women who had lateral chest radiographs and in whom 142 (1.4%) had vertebral fractures reported (22). However, in only 58 (41%) did the presence of a vertebral fracture appear in the final conclusion, in only 23 (16%) was the presence of a vertebral fracture documented in the discharge summary, and only 36% of the patients were using any osteoporosis medications at discharge. The authors concluded that vertebral fractures from lateral chest radiographs represented a missed opportunity for osteoporosis management (22). In a smaller number of women (106) of various ethnicities with a mean age of 65 (range 55e89) yr, the lateral chest radiographs were reviewed with a 1e2-yr follow-up (23). Twenty-six of 106 patients (25%) had vertebral fractures; the fracture prevalence increased with age (in 17 of 54 [13%] women under the age of 65 yr; in 19 of 52 [37%] women older than 65 yr), and 3 of 16 (19%) developed interval fractures. In only 4 of 26 (15%) patients was the fracture included in the report; and although 31 of 106 (29%) were scheduled for bone densitometry, this was performed in only 6 of 106 (6%)
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures
3
Fig. 1. Radiographs: vertebral fractures can opportunistically be identified from any radiographs performed for various clinical indications and which include the spine in the field of view. (A) Anteroposterior (AP) abdominal radiograph showing vertebral fractures of T11 and T12, which are reduced in height and have increased density of the end plates. In such cases, dedicated AP and lateral thoracic and lumbar spinal radiographs should be performed. (B) Lateral lumbar spinal radiograph confirms grade 3 severe vertebral fractures of T11 and T12 and shows additional vertebral fractures of L2, L3, and L4. (C) Lateral chest radiograph with a grade 3 severe wedge fracture of T6 and additional fractures of T11 and T2; the appearances indicate clinical spinal osteoporosis. patients (23). The lateral chest radiograph, particularly in an elderly population, therefore offers an opportunity to identify vertebral fractures, which may be present and which will impact on patient management. However, this will require
the radiologist to actively scrutinize the image for vertebral fractures being present and then report them clearly as fractures. There is evidence that simple training programs improve such vertebral fracture identification (24,25).
Fig. 2. Dual-energy X-ray absorptiometry: (A) Lumbar spine posterior-anterior (PA) bone mineral density scans that show vertebral fractures of T12 and L2, which are reduced in height with sclerosis and concavity of the upper end plates. With such findings, anteroposterior and (B) lateral spinal radiographs should be performed to confirm the presence of fractures; there is a grade 3 severe crush fracture of T12 and a grade 1 mild upper end plate of L2. Alternatively, if appropriate software is available, then (C) DXA lateral vertebral fracture assessment can be performed. In this different patient, there is a grade 3 severe crush fracture of T12. Note aortic calcification anterior to the lumbar spine from which aortic calcification scoring can be derived and which is related to cardiac morbidity and mortality. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
4
Adams
Dual-Energy X-Ray Absorptiometry
Computed Tomography
Vertebral fractures can be evident on the posterior-anterior (PA) DXA scan performed to measure BMD (Fig. 2A). There may be reduction in vertical height and area of the vertebral body affected, and consequently higher BMD, particularly when compared with any previous DXA examinations. With the technical developments in DXA (improved sensitivity of detectors; small increase in X-ray exposure and increased spatial resolution [0.50e0.35 mm]), image quality has improved. This enables better visualization of the vertebral end plate changes, which occur with vertebral fractures. If vertebral fractures are present on the lumbar spine PA BMD scan, then either anteroposterior (AP) and lateral spinal radiographs (Fig. 2B) or DXA VFA (Fig. 2C), if relevant software is available on the scanner, should be performed to confirm the presence of vertebral fracture as this will influence patient management. Alternatively, a protocol could be put in place to perform VFA at the same time as DXA for BMD is performed, which makes scheduling of appointments easier. In our unit, we perform VFA in patients referred for BMD in women aged more than 65 yr, men aged more than 70 yr and all who have had significant (O5 mg/d for longer than 3 mo) oral glucocorticoid therapy over 45 yr, and who have not had any spinal imaging in the previous past 12 mo. Such a protocol can enhance vertebral fracture identification and so improve targeted treatment and management (26).
CT scans are performed widely for a variety of clinical indications. Preceding each CT examination preliminary scan projection radiographs (scout views), generally AP and lateral, are performed to define the anatomical area to be scanned with transverse axial images. The lateral scout image should be carefully scrutinized for vertebral fractures (Fig. 3A). In the Framingham Osteoporosis Study, participants included 50 women and 50 men (age 50e87 yr, mean 70 yr) (27). On lateral CT scout views, T4eL4 vertebrae were assessed independently by 2 radiologists on 2 occasions using an SQ scale as normal, mild, moderate, or severe fracture. The vertebraspecific prevalence of grade 1 (mild) fracture ranged from 3% to 5%.The inter-reader agreement for grade 1 vertebral fractures was fair (k 5 56%e59%) and good (k 5 68% e72%) for grade 2 fractures. Intra-reader agreement for grade 1 vertebral fracture was fair (k 5 55%) for 1 reader and excellent for another (k 5 77%), and intra-reader agreement for grade 2 vertebral fracture was excellent for both readers (k 5 76% and 98%). Thoracic vertebrae were more difficult to evaluate than those in the lumbar region, and agreement was lowest (inter-reader k 5 43%) for fracture at the upper (T4eT9) thoracic levels and highest (inter-reader k 5 76% e78%) for the lumbar spine (L1eL4). The authors concluded that lateral CT scout views can be useful to assess for vertebral fracture (27). In another study of 500 patients (303 males, 197
Fig. 3. Computed tomography: (A) Lateral scout view demonstrating grade 2 moderate upper end plate fracture of L1. (B) Midline sagittal reformation (SR), which can be derived from all MDCT scans with no additional scanning having to be performed or ionizing exposures to the patient, showing grade 3 severe fractures of T3 and T7 in a different patient. Such images are particularly sensitive to identifying vertebral fractures because the middle of the end plate is weaker than the peripheral ring, and will demonstrate fractures not evident on the transverse axial images, so it is essential that such SR images are routinely acquired and always carefully scrutinized for fractures. Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures females, aged 64.6 13.5 yr old) who had CT scans, the identification of vertebral fractures from the CT scout view was compared with CT midline SR (28). In a significant number of scans (59 of 500 [11.8%]), the evaluation on scout CT was incomplete or limited for patient/technical-based conditions (obesity, scoliosis). In 67 of 485 patients (13.8%), 99 vertebral fractures were detected of which only 18 (26.9%) were included in the report. Scrutiny of the scout view took only 41 s and is advocated by the authors to identify vertebral fractures (28). Members of the same group reviewed 300 CT examinations of the thoracic and/or lumbar spine, which were collected and independently analyzed by 3 musculoskeletal radiologists in 2 different sessions (29). A total of 1522 vertebrae were considered (130 males and 170 females; ages, 73.0 2.8 yr). In 73 of 1522 (4.8%), vertebral fractures were identified in 34 of 300 patients (11.3%). The sensitivity and specificity of CT scout views for identification of vertebral fractures were 98.7% and 99.7%, respectively. Accuracy (Area under the Receiver Operator Characteristics [AUROC] 5 0.992 0.008) and interobserver agreement (k 5 0.968 0.008) were excellent. Intraobserver agreement was perfect (k 5 1.000). Performance of this method was independent of spondylosis, vertebral level, and type and grade of vertebral fractures. The authors concluded that CT scout views are a simple but very accurate method for the detection of vertebral fractures (29). With the technical developments in CT (spiral rotation of the X-ray tube; multiple detectors; multidetector spiral CT; MDCT), which have occurred over the past decade, scanning is extremely fast (the torso is imaged in less than 20 s), and spatial resolution improved (0.6 mm) with 3D volumetric or 2D coronal/SR acquired rapidly. These reformatted images require no additional scanning to be performed and involve no additional exposure of the patient to ionizing radiation. The peripheral ring of the vertebral body end plate is the strongest part and the central area the weakest, and it is in this central area where insufficiency microfractures may occur first. This makes midline sagittal CT reformations the most sensitive method for identifying vertebral fractures (Fig. 3B), and which will not be evident on the transverse axial sections. MDCT of the thorax and abdomen is now performed widely worldwide for various clinical indications and is established as probably the most useful method to opportunistically identify vertebral fractures, but again, there is evidence that there is underreporting of these fractures. In an early study of 200 consecutive MDCT with midline SR in patients aged 61 (range 18e92) yr (48% women), there were 70 (35%) with vertebral fractures, of which 51 (73%) were mild grade 1, 13 (19%) were moderate grade 2, and 6 (9%) were severe grade 3 fractures. However, only 6 (9%) were included in the radiology report (30). In a similar study, consecutive MDCT scans performed in 192 patients (95 women) with a mean age of 70.1 yr using slice thickness of 1.25 mm (13) were reviewed. There were 38 (19.8%) patients who had 1 or more grade 2 moderate or grade 3 severe vertebral fractures, but only 5 (11%) were recorded as osteoporotic vertebral fractures in the formal report. The sensitivity of
5 identification of the vertebral fracture on the axial sections was low at 0.35 (13). In another study, abdominal or thoracoabdominal MDCT was performed in 112 postmenopausal women, with axial images and SR being analyzed separately by 2 radiologists. In 27 patients, osteoporotic vertebral fractures were found; only 6 of these could be identified on the axial images, but none of these were included in the radiology report. The authors concluded that SR of standard MDCT images provide important additional information on spinal abnormalities; in particular, osteoporotic vertebral fractures are substantially better detected (31). In a further study, 323 consecutive patients (196 males, 127 females) with a mean age of 62.6 yr (range 20e88) who had chest and/or abdominal MDCT were evaluated (14). Sagittal reformats of the spine obtained from thin section data sets were reviewed by 2 radiologists and assessed for vertebral fractures. A vertebral body height loss of 15% or more was considered a fracture and graded as mild (15%e24%), moderate (25%e49%), or severe (more than 50%). Thirty-one of 323 patients (9.5%) had at least 1 vertebral fracture, and 7 of those patients had multiple fractures with a total of 41 fractures. Morphometric grading revealed 10 mild, 16 moderate, and 15 severe fractures. Prevalence was higher in women (14.1%) than men (6.6%) and increased with patient’s age with 17.1% prevalence in postmenopausal women. Only 6 of 41 vertebral fractures (14.6%) had been noted in the radiology final report, whereas the remaining 35 (85.4%) had not (14). In an audit study, in 175 consecutive patients aged O65 yr, sagittal reformatted images of the spine were obtained from CT scans of the chest or abdomen (32). Vertebral fractures were defined using an SQ technique. The prevalence of vertebral fractures was 13%, with 41 vertebral fractures identified in 22 patients; 12 of 22 (55%) had vertebral fracture included in the formal CT report. The vertebral fracture was newly identified in 17 (77%) patients, but vertebral fracture and osteoporosis were each listed in the relevant discharge summary or clinic letter for only 14% of patients, and only 31% of patients with fracture subsequently received osteoporosis treatment. The authors concluded that sagittal reformatted images of the spine from CT scans of the chest or abdomen detect previously unidentified vertebral fractures and offer an undervalued opportunity to assess fracture risk and intervene with treatments that prevent fractures and reduce mortality. In an in vitro study using 65 vertebrae from 21 human cadavers spines using 64-row multidetector CT scanner, axial images were acquired with slice thicknesses of 0.6, 1, 2, 3, and 5 mm, and SR were obtained using these data sets (33). The specimens were also radiographed in AP and lateral orientation. Vertebrae visualized in the different image data sets were separately graded by 4 radiologists. Fracture status determined in a consensus reading of interactive reformations of the 0.6-mm CT data set in all 3 dimensions served as a standard of reference in combination with pathological examinations. The average agreement for the 0.6-mm SR obtained between each radiologist and standard of reference for the grading of the fractures was very good (k 5 0.81). It was good for the 1-, 2- and 3-mm SR (k 5 0.70, 0.69 and
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
6
Adams
0.64), but only moderate for the radiographs (k 5 0.52), and fair for the 5-mm SR (k 5 0.33). When considering only the detection of fractures, independent of the grading, all kappa values improved by about 0.15, resulting in excellent values for the 0.6 mm through 3-mm SR (0.95 ! k ! 0.79) and good values for the radiographs (k 5 0.72). Ninety-five percent of the fractures could be identified using the 1-mm SR, but 18% of the fractures were missed on the radiographs. The authors concluded that sagittal CT reformations could more accurately assess vertebral fractures than standard spinal radiographs. However, for reliable detection of these fractures, SR derived from axial images with a slice thickness of 3 mm or less are required. The thinnest available axial slice thickness (0.6 mm) performed best in fracture grading, and this is what is generally used in clinical practice (33).
Magnetic Resonance Imaging MRI plays an important role in differentiating acute from long-standing vertebral fractures by revealing the presence of marrow edema in acute fractures and restoration of marrow fat signal in old. MRI is also the imaging method of choice to differentiate vertebral fracture due to benign and malignant etiologies, with the latter having altered soft-tissue signal in the vertebral body, convexity of the posterior vertebral margin, cortical destruction, a soft-tissue mass, and high signal on diffusion-weighted sequences (19). Unlike CT midline sagittal magnetic resonance (MR) images using specific sequences are acquired directly and do not need to be reformatted. Vertebral fractures may be identified opportunistically on MRI scans, which have been performed for various and other clinical indications and include the spine. Before an MR scan, a localizer scan is performed to define the anatomical location of the other scan sequences to be performed. Localizer scans are derived from thick slices in 3 planes (axial, coronal, and sagittal) and so are quite rudimentary images that are limited in spatial resolution and image quality. Despite this poor image quality, there is evidence that such localizer images can be used to identify vertebral fracture. However, subsequently acquired directly sagittal images using specifically selected sequences (Fig. 4) give much higher quality images than the localizer. In a study of 300 MR scans (147 males, 153 females, 59.4 16.4 yr) of the thoracic and/ or lumbar spine, a total of 2186 vertebrae were assessed (34). The initial assessment was by an SQ approach, and morphometric analysis was performed when a vertebral fracture was suspected on the MR localizer images evaluated by 3 radiologists on 2 separate occasions. A full diagnostic sagittal T1weighted fast spin echo MR sequence was used as standard of reference for identification of vertebral fractures. Sixty-seven of 2136 (3.1%) vertebral fractures were identified in 23 of 300 (7.7%) patients. In detection of fractures, sensitivity and specificity of the MR localizer were both 100% (accuracy AUROC 5 1.000). Interobserver agreement was excellent (k 5 0.938 0.013), whereas intraobserver agreement was perfect (k 5 1.000). The diagnostic performance was independent of vertebral level, type, and grade of fractures. In another study, MR localizer sagittal images of 856 patients
Fig. 4. Magnetic resonance imaging: although vertebral fractures can be identified on the initial localizer views, these are of limited image quality. Subsequently acquired direct sagittal images using specifically selected sequences give much higher quality images than the localizer. T2-weighted sagittal image confirms grade 3 severe wedge fractures of T7 and T9; the latter shows high signal in the marrow indicating edema and that the fracture is recent. The fracture at T9 has regained normal marrow fat signal indicating that it is an old fracture. undergoing breast MRI were reviewed by 3 expert musculoskeletal radiologists with an SQ approach to detecting vertebral fractures (35). Fifty-seven MR localizer images (57 of 856; 6.7%) were inadequate for diagnostic purposes. MR localizer images detected vertebral fractures in 71 of 799 patients (8.9%); of these, 63 of 71 (88.7%) vertebral fractures were not reported. The authors concluded that MR localizer images should always be scrutinized for the presence of vertebral fractures in an MRI scan being performed for other clinical indications.
Radionuclide Imaging RNS with technetium (99mTc) bisphosphonates is a good skeletal survey technique to identify areas of abnormal uptake of isotope, which generally reflects increased blood flow, bone turnover, and bone formation. The technique is very sensitive, but not specific in that numerous pathologies (osteoarthritis, osteomyelitis, blastic bone metastases, Paget’s disease of bone, fractures, Looser zones) result in increased uptake.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures Acute vertebral fractures will have increased uptake, which is usually apparent within 48 h of fracture and will persist for between 6 and 18 mo (36,37; Fig. 5). A fracture may therefore be an incidental finding in a patient undergoing RNS for other clinical indications, for example, for bone metastases in known primary cancer. However, having cancer does not preclude the patient having insufficiency fractures and/ or osteoporosis. Positron emission tomography CT with F18 fluorodeoxyglucose is increasingly used in patients with a variety of cancers for tumor staging. There may be increased uptake of radionuclide in osteoporotic fractures, which must therefore be differentiated from bone metastases (38).
Discussion From the peer-reviewed literature, there is good evidence that many imaging techniques being performed for various clinical indications offer the opportunity to identify vertebral fractures incidentally if the spine is in the field of view. These include radiographs (10e12,21e23), CT scout views and SR (13,14,27e33), MRI localizers, and sagittal images using specific sequences (19,34,35) and radionuclide bone scans (36e38). Probably the most sensitive is midline sagittal CT reformations, which are very rapidly (a few seconds) acquired by technical staff at the time of the scan, and this should be routine practice in radiology departments. This is more efficient and preferable to leaving the choice to obtain and perform such reformats to the radiologist at the time of reporting. Such midline SR will identify vertebral fractures, which are not evident on transverse axial images (13,31), so, it is essential for these reformatted images to be viewed and scrutinized at the time of reporting, otherwise vertebral fractures will be missed. MDCT is now very widely performed, especially in elderly patients in whom the prevalence of vertebral
Fig. 5. Radionuclide bone scan: such a scan provides a good survey technique of the whole skeleton is sensitive but not very specific and is performed for a variety of clinical indications. PA image: there is increased uptake in the body of T7 due to an acute vertebral fracture. Such fractures show increased uptake within 24 h of the fracture occurring, but the increased uptake may persist for up to 2 yr.
7 fractures will be higher than in a younger cohort. However, although vertebral fractures are present on an image, there is evidence that there is widespread and worldwide underreporting in all imaging techniques (10e15). In a review of the literature in 2010 from 12 studies (7 using radiographs; 5 using MDCT), reporting rates were low with a mean value of 27.4% (range 0%e66.3%) and were significantly lower in MDCT than in radiographs (mean 8.1% vs 41.2%; 12). So, although scientific studies have shown a high prevalence of opportunistically imaged vertebral fractures, the very significant underreporting of such fractures will have a detrimental effect on patient management. This underreporting occurs despite the vertebral fracture initiative of the International Osteoporosis Foundation having been launched over a decade ago. The problem is largely that ‘‘observers will only see what they look for’’ and spine images are not being scrutinized by radiologists. This is likely to be due to them reporting CT scans in their particular area of special expertise (gastrointestinal, cardiothoracic, and so forth) and so being unaware of the importance and relevance of vertebral fractures and their significance to the appropriate management of patients with osteoporosis. Constant and reiterative teaching and training of radiologists to emphasize the importance of reporting vertebral fractures is required (24,25). Another potential solution might be the use of automated computeraided diagnostic tools to prompt the radiologist to recognize that vertebral fractures may be present in the image being reviewed. A commercial tool is available (SpineAnalyzer, Optasia Medical, Cheadle, UK), which provides a semiautomated quantitative vertebral morphometry devised from shape-based statistical modeling and has been applied in several studies (39e41). In 1 report, based on the Framingham Heart Study Offspring and Third Generation MultiDetector CT Study (39), 1246 vertebrae were analyzed on lateral CT scout views in 96 subjects, and the authors concluded that the semiautomated algorithm provided excellent intra- and inter-reader reliability for vertebral height measurements, along with good to fair reliability for vertebral height ratios. Reliability for VFA based solely on quantitative morphometry was also good and was comparable to previous reports for SQ vertebral fracture grading by radiologists (39). In a report from the population-based Rotterdam study, which includes 14,926 inhabitants aged 45 yr and older in a subset of 99 lateral spinal radiographs, Cobb’s kyphosis angle was calculated manually and using SpineAnalyzer. The authors conclude that such vertebral fracture morphometric data to derive the Cobb’s kyphosis angle are relatively reliable, and that additional data (vertebral wedging, intervertebral disc space, spondylolisthesis, and the lordosis angle) can be derived. The average time to complete the semiautomated morphometry analysis was approximately 9 min 40 s less than previously reported for manual morphometry analysis (40), and so may be particularly useful in application to large osteoporosis epidemiological or therapeutic efficacy studies. SpineAnalyzer has also been applied in the Mr Os study (41), a cohort of community-dwelling men aged 65 yr in whom lateral spinal radiographs were available in 5958 of
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
8
Adams
5994 participants at visit 1, and 4399 of 4423 participants at Visit 2 after 4.6 yr. A subset of 494 subjects’ images was read manually to assess the sensitivity and specificity of triage by the semiautomated tool. Triage identified 3215 (53.9%) participants with radiographs that required further evaluation by the physician reader. For prevalent fractures at visit 1 (SQ 1), intra-reader kappa statistics ranged from 0.79 to 0.92; percent agreement ranged from 96.9% to 98.9%; sensitivity of the triage was 96.8%, and specificity of triage was 46.3%. The authors concluded that this triage process reduced expert reader workload without hindering the ability to identify vertebral fractures (41). Other groups have reported semiautomated tools for vertebral fracture identification in MDCT (42), in DXA VFA images (43), and spinal radiographs (44), which show potential. With further developments in computer vision techniques (45), it may become feasible to implement a completely automated search of a radiology picture archiving and communication systems (PACS) for images that contain vertebral fracture, which would be ideal. There is evidence that the presence of incidentally identified and reported vertebral fractures improves rates of bone densitometry performed and treatments implemented by clinicians (46), and that such a strategy is cost-effective (47). In conclusion there is a great potential to identify vertebral fractures opportunistically using a variety of imaging techniques, which include the spine in the field of view and which are being performed for other and varied clinical reasons. This is beneficial to patient management, is cost effective, and so has the potential to reduce future fracture risk. Unfortunately, however, there are numerous reports of the underreporting of such fractures by radiologists. Educational and training initiatives are beneficial to all clinicians, but constant reiteration is required to reinforce the message of the importance and relevance of accurate and clear reporting of vertebral fractures, particularly by radiologists, to determine appropriate patient management. To this end, it is essential that radiologists always scrutinize any images of the spine for vertebral fracture, as what is not sought will not be evident. Semiautomated/automated computer tools show potential to improve the identification of vertebral fractures and reduce the amount of manual interpretation required in large epidemiological and therapeutic efficacy trials by using such tools in a triage protocol.
Acknowledgments The author has nothing to disclose.
References 1. Christiansen BA, Bouxsein ML. 2010 Biomechanics of vertebral fractures and the vertebral fracture cascade. Curr Osteoporos Rep 8(4):198e204. 2. Delmas PD, Genant HK, Crans GG, et al. 2003 Severity of prevalent vertebral fractures and the risk of subsequent vertebral and nonvertebral fractures: results from the MORE trial. Bone 33(4): 522e532.
3. Klotzbuecher CM, Ross PD, Landsman PB, et al. 2000 Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 15(4):721e739. 4. Lindsay R, Silverman SL, Cooper C, et al. 2001 Risk of new vertebral fracture in the year following a fracture. JAMA 285(3):320e323. 5. Crandall CJ, Newberry SJ, Diamant A, et al. 2014 Comparative effectiveness of pharmacologic treatments to prevent fractures: an updated systematic review. Ann Intern Med 161(10): 711e723. 6. Reid IR. 2015 Efficacy, effectiveness and side effects of medications used to prevent fractures. J Intern Med 277(6):690e706. 7. Johnell O, Kanis JA. 2006 An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 17(12):1726e1733. 8. Kanis JA, Oden A, Johansson H, et al. 2009 FRAX and its applications to clinical practice. Bone 44(5):734e743. 9. Ensrud KE. 2013 Epidemiology of fracture risk with advancing age. J Gerontol A Biol Sci Med Sci 68(10):1236e1242. 10. Gehlbach SH, Bigelow C, Heimisdottir M, et al. 2000 Recognition of vertebral fracture in a clinical setting. Osteoporos Int 11(7):577e582. 11. Delmas PD, van de Langerijt L, Watts NB, et al. 2005 Underdiagnosis of vertebral fractures is a worldwide problem: the IMPACT study. J Bone Miner Res 20(4):557e563. 12. Bartalena T, Rinaldi MF, Modolon C, et al. 2010 Incidental vertebral compression fractures in imaging studies: lessons not learned by radiologists. World J Radiol 2(10):399e404. 13. Williams AL, Al-Busaidi A, Sparrow PJ, et al. 2009 Underreporting of osteoporotic vertebral fractures on computed tomography. Eur J Radiol 69(1):179e183. 14. Bartalena T, Giannelli G, Rinaldi MF, et al. 2009 Prevalence of thoracolumbar vertebral fractures on multidetector CT: underreporting by radiologists. Eur J Radiol 69(3):555e559. 15. Lentle BC, Brown JP, Khan A, et al. 2007 Recognizing and reporting vertebral fractures: reducing the risk of future osteoporotic fractures. Can Assoc Radiol J 58(1):27e36. 16. Genant HK, Wu CY, van Kuijk C, Nevitt MC. 1993 Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 8(9):1137e1148. 17. Guglielmi G, Diacinti D, van Kuijk C, et al. 2008 Vertebral morphometry: current methods and recent advances. Eur Radiol 18(7):1484e1496. 18. Jiang G, Eastell R, Barrington NA, Ferrar L. 2004 Comparison of methods for the visual identification of prevalent vertebral fracture in osteoporosis. Osteoporos Int 15(11):887e896. 19. Link TM, Guglielmi G, van Kuijk C, Adams JE. 2005 Radiologic assessment of osteoporotic vertebral fractures: diagnostic and prognostic implications. Eur Radiol 15(8):1521e1532. 20. Cauley JA, Zmuda JM, Wisniewski SR, et al. 2004 Bone mineral density and prevalent vertebral fractures in men and women. Osteoporos Int 15(1):32e37. 21. Kim N, Rowe BH, Raymond G, et al. 2004 Underreporting of vertebral fractures on routine chest radiography. AJR Am J Roentgenol 182(2):297e300. J Gen Intern Med. 2006 Apr; 21(4):352-356. 22. Morris CA, Carrino JA, Lang P, Solomon DH. 2006 Incidental vertebral fractures on chest radiographs. Recognition, documentation, and treatment. J Gen Intern Med 21(4):352e356. 23. Mui LW, Haramati LB, Alterman DD, et al. 2003 Evaluation of vertebral fractures on lateral chest radiographs of inner-city postmenopausal women. Calcif Tissue Int 73(6):550e554. 24. Casez P, Uebelhart B, Gaspoz JM, et al. 2006 Targeted education improves the very low recognition of vertebral fractures and
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015
Identification of Vertebral Fractures
25. 26.
27.
28. 29. 30. 31.
32. 33. 34. 35.
osteoporosis management by general internists. Osteoporos Int 17(7):965e970. Gruber M, Dinges J, M€uller D, et al. 2013 Impact of specific training in detecting osteoporotic vertebral fractures on routine chest radiographs. Rofo 185(11):1074e1080. Kuet KP, Charlesworth D, Peel NF. 2013 Vertebral fracture assessment scans enhance targeting of investigations and treatment within a fracture risk assessment pathway. Osteoporos Int 24(3):1007e1014. Samelson EJ, Christiansen BA, Demissie S, et al. 2011 Reliability of vertebral fracture assessment using multidetector CT lateral scout views: the Framingham Osteoporosis Study. Osteoporos Int 22(4):1123e1131. Bazzocchi A, Spinnato P, Albisinni U, et al. 2012 A careful evaluation of scout CT lateral radiograph may prevent unreported vertebral fractures. Eur J Radiol 81(9):2353e2357. Bazzocchi A, Fuzzi F, Garzillo G, et al. 2013 Reliability and accuracy of scout CT in the detection of vertebral fractures. Br J Radiol 86(1032):20130373. Woo EK, Mansoubi H, Alyas F. 2008 Incidental vertebral fractures on multidetector CT images of the chest: prevalence and recognition. Clin Radiol 63(2):160e164. M€uller D, Bauer JS, Zeile M, et al. 2008 Significance of sagittal reformations in routine thoracic and abdominal multislice CT studies for detecting osteoporotic fractures and other spine abnormalities. Eur Radiol 18(8):1696e1702. Chan PL, Reddy T, Milne D, Bolland MJ. 2012 Incidental vertebral fractures on computed tomography. N Z Med J 125(1350): 45e50. Bauer JS, M€uller D, Ambekar A, et al. 2006 Detection of osteoporotic vertebral fractures using multidetector CT. Osteoporos Int 17(4):608e615. Bazzocchi A, Garzillo G, Fuzzi F, et al. 2014 Localizer sequences of magnetic resonance imaging accurately identify osteoporotic vertebral fractures. Bone 61:158e163. Bazzocchi A, Spinnato P, Garzillo G, et al. 2012 Detection of incidental vertebral fractures in breast imaging: the potential role of MR localisers. Eur Radiol 22(12):2617e2623.
9 36. Matin P. 1979 The appearance of bone scans following fractures, including immediate and long-term studies. J Nucl Med 20(12): 1227e1231. 37. Ryan PJ, Fogelman I. 1994 Osteoporotic vertebral fractures: diagnosis with radiography and bone scintigraphy. Radiology 190(3):669e672. 38. Katzel JA, Heiba SI. 2005 PET/CT F-18 FDG scan accurately identifies osteoporotic fractures in a patient with known metastatic colorectal cancer. Clin Nucl Med 30(10):651e654. 39. Kim YM, Demissie S, Eisenberg R, et al. 2011 Intra-and inter-reader reliability of semi-automated quantitative morphometry measurements and vertebral fracture assessment using lateral scout views from computed tomography. Osteoporos Int 22(10):2677e2688. 40. Oei L, Ly F, El Saddy S, et al. 2013 Multi-functionality of computer-aided quantitative vertebral fracture morphometry analyses. Quant Imaging Med Surg 3(5):249e255. 41. Cawthon PM, Haslam J, Fullman R, et al. 2014 Methods and reliability of radiographic vertebral fracture detection in older men: the osteoporotic fractures in men study. Bone 67:152e155. 42. Baum T, Bauer JS, Klinder T, et al. 2014 Automatic detection of osteoporotic vertebral fractures in routine thoracic and abdominal MDCT. Eur Radiol 24(4):872e880. 43. Roberts MG, Pacheco EM, Mohankumar R, et al. 2010 Detection of vertebral fractures in DXA VFA images using statistical models of appearance and a semi-automatic segmentation. Osteoporos Int 21(12):2037e2046. 44. Roberts MG, Oh T, Pacheco EM, et al. 2012 Semi-automatic determination of detailed vertebral shape from lumbar radiographs using active appearance models. Osteoporos Int 23(2):655e664. 45. Roberts MG, Cootes TF, Adams JE. 2012 Automatic location of vertebrae on DXA images using random forest regression. Med Image Comput Comput Assist Interv 15(Pt 3):361e368. 46. Majumdar SR, McAlister FA, Johnson JA, et al. 2012 Interventions to increase osteoporosis treatment in patients with ‘incidentally’ detected vertebral fractures. Am J Med 125(9):929e936. 47. Majumdar SR, Lier DA, McAlister FA, et al. 2013 Costeffectiveness of osteoporosis interventions for ‘incidental’; vertebral fractures. Am J Med 126(2):169.e9e169.e17.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Volume
-,
2015