Clinic Rev Bone Miner Metab DOI 10.1007/s12018-017-9235-7
REVIEW PAPER
Opportunistic Screening for Osteoporosis Using Body CT Scans Obtained for Other Indications: the UW Experience Scott J. Lee 1 & Perry J. Pickhardt 1,2
# Springer Science+Business Media, LLC 2017
Abstract Low bone mineral density (osteoporosis and osteopenia) leading to fragility fractures is associated with significant morbidity and mortality in our aging population. This condition is grossly underdiagnosed due to both insufficient screening and its silent nature prior to complicating fragility fractures. Body CT scans are commonly obtained among older adults for a wide variety of indications and contain rich data regarding bone health that are often ignored. At the University of Wisconsin, we have sought to harness this CT information for Bopportunistic^ osteoporosis screening. In this article, we review the various CT-based approaches we have taken to date, including routine assessment of the spine for both vertebral fractures and trabecular density, as well as assessment of the hip, deriving femoral neck T-scores that are essentially equivalent to dual-energy x-ray absorptiometry (DXA). Future directions of research and clinical implementation are also discussed. Keywords CT . Computed tomography . Opportunistic screening . Osteoporosis
Introduction Osteoporosis and osteopenia are common disease conditions that affect bone strength in adults worldwide. Osteoporosis * Perry J. Pickhardt
[email protected] 1
University of Wisconsin School of Medicine and Public Health, 600 Highland Avenue, Madison, WI 53705, USA
2
Department of Radiology, University of Wisconsin School of Medicine and Public Health, E3/311 Clinical Science Center, 600 Highland Ave, Madison, WI 53792-3252, USA
and osteopenia (collectively termed low bone mineral density (BMD)) confer increased fracture risk [1], and these fractures are associated with significant morbidity and mortality [2]. All types of fragility fractures (which are fractures due to lowenergy mechanisms) have been shown to reduce quality of life [3] and are at least partially preventable if patients with low BMD are effectively identified and treated [4]. Low BMD is asymptomatic until a fragility fracture occurs, which may contribute to under-diagnosis. Screening for osteoporosis also takes additional time and cost on the part of the patient and provider team. Of the available options, dual-energy x-ray absorptiometry (DXA) is the most widely used screening tool for low BMD. However, it is generally underutilized in populations eligible for screening, and screening rates have been shown to vary from 13 to 56% in studies of Medicare claims data and primary care practices [5, 6]. Given the public health burden of low BMD, improving screening rates is of great importance.
CT for BOpportunistic^ Osteoporosis Screening In 2006, an estimated 31.6 million computed tomography (CT) scans that included the chest or abdomen were performed in the USA for a variety of indications [7], all of which include some portion of the lumbar spine [8]. In addition, CT scans of the pelvis image often include the femoral necks, which is discussed later on in this review. In order to maximize the use of BMD data already gathered for often unrelated purposes, we have proposed harnessing these CT images for the valuable information regarding bone mineral density. We utilize this bone data for osteoporosis screening through the simple use of calibrated attenuation values of the trabecular bone, which can be obtained either prospectively or retrospectively from the vast majority of CT scans seen in routine
Clinic Rev Bone Miner Metab
Fig. 1 Transverse (axial) CT images at the L1 vertebral level in four different patients (a–d) viewed in standard soft tissue (row 1) and bone (row 2) window settings. Trabecular bone CT attenuation values are shown in red for each oval region of interest; note that the attenuation measure (in HU) does not change according to the CT window for
viewing. The four patients represent sample BMDs ranging from low (osteoporosis) (a) to high (normal) (d), which is actually more visually apparent on the soft tissue window setting (row 1). From reference [9] with permission
clinical practice. We have termed this process of gathering indirect bone density data from routine CT scans as Bopportunistic osteoporosis screening.^ In essence, the
benefits of opportunistic CT screening are twofold: (1) unsuspected vertebral fractures can be identified and establish a diagnosis of osteoporosis in the appropriate clinical setting
Fig. 2 a Mean trabecular CT attenuation values (SDs) at each vertebral level from T12-L5, stratified by osteoporosis, osteopenia, and normal bone mineral density according to the DXA reference standard. The differences between mean attenuation for each bone mineral density group at each level are significant (p < 0.001). On average, CT attenuation tends to be lowest at the L3 level and increases slightly at higher and lower levels. b The ROC AUCs for osteoporosis are similar at
each individual vertebral level, using a multilevel T12 to L5 average, and in a multivariable model incorporating measures from all levels simultaneously, with broad overlap of 95% CIs. Osteopenia was considered a false-positive result for these calculations. The total number of assessable CT measurements per level was 2016 for T12, 2040 for L1, 2048 for L2, 2046 for L3, 2021 for L4, and 1943 for L5. From reference [9] with permission
Clinic Rev Bone Miner Metab
and (2) these CT scans can provide BMD information both indirectly via vertebral trabecular attenuation values (in HU) and DXA-equivalent femoral neck T-scores. This review will outline the various studies carried out at our institution and others that have explored the use of CT-based spinal trabecular and femoral neck bone assessment for the purpose of opportunistic osteoporosis screening. Performing opportunistic osteoporosis screening in conjunction with body CT interpretation is quite simple and takes little additional training. For rapid spinal assessment, an ovoid region-of-interest (ROI) is placed within the anterior trabecular space of the first lumbar (L1) vertebral body (Fig. 1). Care should be taken to avoid inclusion of cortical bone, focal lesions, or other defects. The L1 level was chosen because it is readily identified as the first non-rib bearing vertebra on transverse (axial) slices and is present on most abdominal and thoracic CT scans. In practice, any vertebral level from T12 through L5 could be employed (Fig. 2), which is useful for cases where a compression deformity involves L1. Placement of the ROI in the anterior trabecular space is important since the posterior trabecular space often contains a greater proportion of non-osseous structures, such as the vertebral venous plexus [10]. These non-osseous posterior structures can lead to false values that are not representative of the majority of the trabecular space. In practice, we typically place the L1 trabecular ROI on the sagittal reconstructions (Fig. 3), which yields Fig. 3 Transverse (axial) and sagittal L1 trabecular attenuation measurement at abdominal CT in a 64-year-old man. Axial (a, b) and sagittal (c, d) CT images in soft tissue (a, d) and bone (c, d) windows show ROI placement in the anterior trabecular space of L1. In addition to low BMD (83 HU axial, 85 HU sagittal) that is likely compatible with osteoporosis, moderate compression fractures are present at T12 and L4 (arrows), indicative of complicated osteoporosis. In retrospect, the patient had a similar L1 attenuation (85 HU sagittal) compatible with osteoporosis at CT 3 years earlier (e) but without complicating fracture at that time. From reference [11] with permission
attenuation values similar to the transverse (axial) view [11]. Regardless of whether or not opportunistic osteoporosis screening is being performed, the sagittal reconstructions should be routinely reviewed for the presence of vertebral compression fractures, which often went undiagnosed prior to these reconstructions becoming standard of care [12]. In our experience, more than 80% of moderate or severe vertebral compression fractures were missed at transverse (axial) interpretation of routine abdominal CT scans before the routine inclusion of the sagittal series (Fig. 4). Our initial study showing that simple and rapid CT-based attenuation measurements are able to identify patients with DXA-proven osteoporosis was performed using 252 patients who underwent CT colonography and DXA within a 6-month period [13]. We found that a trabecular attenuation threshold of 160 Hounsfield units (HUs) was 100% sensitive for the diagnosis of DXA-proven osteoporosis, albeit fairly nonspecific. The reference standard DXA value used was the lowest central T-score measurement value between the lumbar spine and the femoral neck, as might be used in clinical practice. This initial study also found that the simple ROI method was equally effective to the Bphantomless^ quantitative approach, which was much more onerous. The diagnostic performance of this CT-based opportunistic osteoporosis screening was then expanded upon in another study of 1867 adults who underwent both abdominal CT and
Clinic Rev Bone Miner Metab Fig. 4 CT images obtained in a 90-year-old woman presenting with flank pain but without apparent cause identified on the prospective CT report. Left: Transverse images obtained through the superior (top), middle (middle), and inferior (bottom) portions of the compression fracture show that the fracture may have been mistaken for degenerative changes. Right: Retrospective sagittal reconstruction reveals a severe T12 wedge compression fracture (arrow). From reference [12] with permission
central DXA within a 6-month period at our institution [9]. There were a wide variety of clinical indications for CT scanning, such as abdominal pain and oncologic follow-up. The median time between CT and DXA studies was 67 days (interquartile range 27 to 188 days). The resulting diagnostic performance statistics are summarized in Table 1. Three separate L1 trabecular attenuation thresholds (160, 135, and 110 HU) were proposed that provided varying targets for sensitivity and specificity. For example, a threshold of ≤ 160 HU was highly sensitive (90.0%) but had reduced specificity (52.3%), whereas a threshold of ≤ 110 HU was highly specific (91.3%) but had reduced sensitivity (52.1%). An additional interesting and important observation was that 97% of patients with a significant vertebral compression fractures had an L1 trabecular attenuation ≤ 145 HU, yet had non-osteoporotic central DXA T-scores (Fig. 5). Two independent external validation studies of the diagnostic performance of this CT-based opportunistic osteoporosis screening approach were carried out [14, 15] and found
agreement between CT attenuation values and DXA-proven osteoporosis, with similar sensitivity but lower specificity values at a 160 HU cutoff (91–91.7% and 27–29%, respectively). A third study found only a weakly positive correlation between L1 HU and DXA [16]. The differences in diagnostic performance between these studies may reflect differences in imaging techniques and local patient populations. It should also be noted that the DXA reference standard represents an imperfect comparator, especially for spinal assessment. Given these results, we suggest that the more sensitive threshold of 160 HU (~ 90%) be used as a Brule-out^ test, where individuals with an L1 attenuation exceeding this value be considered at low risk for osteoporosis; this notion is further supported by the negative predictive value of 94.7% for individuals above the 160 HU threshold found by Pickhardt et al. [9]. Conversely, the more specific threshold of ≤ 110 HU (~ 90%) should be used as a Brule-in^ test, whereby individuals with L1 attenuation below this value should be considered for further confirmatory BMD testing. This 110 HU
Clinic Rev Bone Miner Metab Table 1 Diagnostic performance statistics of L1 CT attenuation values for distinguishing osteoporosis from nonosteoporosis using DXA reference standarda
Variable
High sensitivity threshold
Balanced threshold
High specificity threshold
L1 CT attenuation, HU
≤ 160
≤ 135
≤ 110
Sensitivity Specificity
90.0 (86.9–92.4) 52.3 (49.8–54.8)
75.5 (71.4–79.2) 75.4 (73.2–77.4)
52.1 (47.5–56.6) 91.3 (89–8 – 92.6)
PPV
35.5 (32.8–38.3)
47.2 (43.6–50.8)
63.5 (58.5–68.2)
NPV
94.7 (93.0–96.0)
91.3 (89.7–92.7)
86.7 (85.0–88.2)
Positive likelihood ratio Negative likelihood ratio
1.89 (1.78–2.00) 0.19 (0.14–0.25)
3.06 (2.77–3.39) 0.33 (0.28–0.38)
5.96 (4.97–7.14) 0.53 (0.48–0.58)
All values other than L1 attenuation and likelihood ratios are given as B% (95% Confidence Interval)^ CT computed tomography, DXA dual-energy x-ray absorptiometry, NPV negative predictive value, PPV positive predictive value a
Adapted from Pickhardt et al., Opportunistic screening for osteoporosis using abdominal computed tomography scans obtained for other indications [9]
threshold had a positive likelihood ratio of 5.96 and a positive predictive value of 63.5% [9]. If a clinician wishes to be more conservative and avoid Bincidentaloporosis^ [17], an even lower and more specific threshold such as ≤ 90 HU could be used before making a recommendation for further testing, since we have shown that patients below this threshold are
Fig. 5 The black dots represent patients with CT-detected T12 to L5 compression fractures (not shown here are 15 patients with L1 compression fractures where reliable CT-attenuation measurement at this level was not possible). Note the broad range of DXA T-scores among patients with vertebral fracture, including normal scores. Overall, more than half of all patients with fractures had a nonosteoporotic DXA T-score, but 97% had an L1 or mean vertebral attenuation ≤ 145 HU. From reference [9] with permission
at greatly increased risk for both prevalent and future incident fragility fractures [18, 19]. The inter- and intra-observer agreement of L1 attenuation measurements has been assessed in multiple studies and shown to be quite excellent. Pickhardt et al. found the Bland-Altman 95% limits of agreement to be − 6.1 to 16.3 HU when two readers measured the combined mean attenuation for T12-L5 [13]. More importantly, there were no diagnostic reclassifications for osteoporosis when using a threshold of 160 HU. Pompe et al. [20] found an intra-class correlation coefficient of 0.92 (95% CI 0.90–0.94) for measurements of L1 attenuation. A third independent study showed intra-class correlation coefficients ranging from 0.94 to 0.99 (95% limits of agreement, − 21.9 to 20.6 HU) for L1 attenuation [21]. It is important to be aware of certain CT image acquisition parameters that affect the practice of opportunistic osteoporosis screening. For example, the administration of IV contrast material tends to increase L1 trabecular attenuation. Although Pickhardt et al. [9] did not directly look at attenuation differences due to IV contrast material, DXA-proven osteoporosis diagnostic performance did not significantly differ with and without its use (non-contrast AUC = 0.83; contrast-enhanced AUC = 0.84). Pompe et al. [22] found a median L1 attenuation increase of 16–19 HU between unenhanced and portal venous phase images and suggested using phase-specific diagnostic thresholds to avoid false-negative results. They also found that L1 attenuation on venous-phase images was slightly increased relative to arterial-phase images (median increase, 4.0 HU). A later study by Pickhardt et al. [23] employing a similar design found a mean L1 attenuation difference of 11 HU between unenhanced and portal-venous phase images. The accuracy of L1 attenuation in diagnosing low BMD again did not significantly differ when between contrast-enhanced and unenhanced images. More importantly than contrast effects, there is a marked attenuation change in osseous structures that occurs at different single-energy CT voltages, which
Clinic Rev Bone Miner Metab
precludes the use of the above-outlined 120 kVp-based HU thresholds for scans acquired at other voltages, such as 100 or 140 kVp [24]. As an example, Garner et al. showed that mean L1 attenuation at 80 kVp to be 76 HU (65%) higher than that noted at 140 kVp. Further experimental research and phantom studies that quantify acquisition effects and the development of software that automatically adjusts bone attenuation measurements to the well-studies 120 kVp reference values would expand the number of CT image sets in which opportunistic osteoporosis screening could be performed. A final note of caution is that CT scans from different vendors may also yield slightly different attenuation values for the same anatomic structure in the same patient [25]. Given the demonstrated effects of IV contrast, x-ray tube voltage, and CT manufacturer on L1 attenuation, we do not recommend using CT-based L1 HU opportunistic osteoporosis screening to formally replace DXA screening at this time. Rather, in our practice, we use this simple opportunistic screening action to trigger more dedicated consideration of fracture risk (e.g., clinical FRAX assessment), especially if patients have not had prior screening.
Asynchronously Calibrated QCT Quantitative computed tomography (QCT) is an accurate method for measuring BMD that has some distinct advantages over DXA, especially for the spine [26]. QCT can directly measure trabecular BMD (without the cortical overlay); is more sensitive to early trabecular bone loss in the lumbar spine [27]; and can provide data for more advanced methods of mechanical bone strength assessment, such as finite element analysis (FEA). However, QCT use has been limited
by efforts to minimize lifetime patient radiation exposure, access time, and costs for general purpose CT scanners and dedicated staff to perform synchronous phantom calibration. More recently, technological developments have circumvented the need for synchronous phantom calibration [28] and now allows for BMD assessment on conventional CT scans obtained for any other routine clinical indication. This method is called asynchronous QCT and disconnects the need for simultaneous scanning of the phantom and patient. Asynchronous QCT is not to be confused with truly phantomless approaches that use internal controls such as patient muscle and fat [13]. Asynchronous QCT uses phantom data obtained from quality assurance (QA) scans when a patient is not present, while traditional QCT must have an appropriate BMD phantom in the scan field concurrently with the patient. The BMD calibration equation is obtained from the QA scan and used to convert attenuation values into BMD measurements on conventional CT scans, either retrospectively or prospectively. Asynchronous QCT has been shown to be a reliable method for measuring BMD. Brown et al. compared conventional and asynchronously calibrated QA scans and found no significant difference in mean spinal volumetric BMD (vBMD) and actually found a lower standard deviation with the asynchronous method [28]. In addition to vBMD, asynchronous QCT can measure areal BMD (aBMD) as well. Brown et al. found no significant difference in the mean or variance of femoral neck aBMD between synchronous and asynchronous QCT [28]. Furthermore, aBMD can be converted into widely recognized T- and Z-scores from conventional CT data. This process of using a QCT approach to derive DXA-equivalent femoral neck T-scores is sometimes referred to as BCTXA,^ although not truly a formal acronym but rather an allusion to DXA (Fig. 6).
Fig. 6 a CT image through pelvis from screening colonography exam. The green box is utilized to segment the hip for automated analysis. b CT-derived CTXA image. Green box defines the femoral neck for BMD determination. From reference [29] with permission
Clinic Rev Bone Miner Metab
CTXA and DXA T-scores of the femoral neck are highly correlated. In a study of female patients who underwent CT colonography followed by DXA within 9 months, we found that CTXA and traditional DXA femoral neck T-scores had an r2 = 0.907, even though time gaps of up to 4 years took place between patient CT data acquisition and QA calibration for asynchronous QCT [30]. This gap between CT and asynchronous calibration may in part account for the + 0.3 T-score bias by DXA relative to asynchronous QCT. In a separate study at our institution by Ziemlewicz et al. [31], matched femoral neck CTXA and DXA T-scores in 355 adults had an r2 = 0.82 (Fig. 7), and a small positive T-score bias of DXA relative to CTXA was again observed (+ 0.18). Based on the results of these studies, it might be reasonable to add 0.2–0.3 units to femoral neck CTXA to correct for the positive DXA bias, but this may not be necessary in a general-purpose screening environment where the goal is to opportunistically identify high-risk, unscreened patients at CT who might not otherwise undergo DXA screening. This would especially be true in patients with severely osteoporotic CTXA Tscores, where the potential bias between CTXA and DXA femoral neck T-scores would become moot and the importance of dedicated BMD follow-up and possible treatment should be emphasized. The same argument applies to the effect of IV contrast on CTXA T-scores. Ziemlewicz et al. directly compared unenhanced and IV contrast-enhanced CTXA studies of the femoral neck in 410 adult patients undergoing CT urography and found a small constant offset correction of nearly 0.3 T-score units [32]. In essence, the CTXA and DXA T-scores can be considered equivalent for the purposes of opportunistic screening. Fig. 7 Linear regression analysis of QCT (CTXA) and DXA Tscore stratified by gender. From reference [30] with permission
Future Directions Fully automated trabecular spinal segmentation and measurement algorithms could reduce or eliminate inter- and intraobserver variation and improve the precision of opportunistic osteoporosis screening. Investigational algorithms have been implemented using CT colonography screening studies with promising results (Fig. 8) [29]. Automated measurements of trabecular attenuation could also help establish populationlevel reference standards through the collection of large datasets. CT colonography screening is an especially attractive target for early implementation of CT-based osteoporosis screening due to the preventive goals of the exam and the general homogeneity of image acquisition parameters. We currently provide free opportunistic spinal and femoral neck screening at CT colonography, including fracture assessment and hip CTXA T-scores [33]. Because the CTXA T-scores are considered DXA-equivalent, this screening Bcounts^ in terms of the patient’s health maintenance record. A similar approach would also be useful in the setting of CT-based lung cancer screening, and indeed, the use of attenuation values to has been shown to detect concomitant low BMD in regular tobacco smokers [34]. More broadly, we are working on ways to prospectively institute the option of adding opportunistic osteoporosis screening to eligible older adults undergoing body CT scanning for any indication via the electronic ordering system. Finite element analysis (FEA) is a widely used simulation method in the field of engineering that virtually applies mechanical stresses to a mathematical representation of a structure. It has been applied to spine and hip structures to quantitate structural strength. Until recently, the use of FEA in CT has remained limited to the QCT setting, but it is now possible
Clinic Rev Bone Miner Metab Fig. 8 Fully-automated CT bone densitometry calculation. Computer software identifies the spine, locates the lowermost (12th) rib, and identifies the L1 and L2 vertebral bodies (lower two lines). The software next segments the vertebral body. The segmented vertebral body is then eroded by 5 mm to remove the cortical bone and retain the medullary bone (green shaded, lower image of an L2 vertebral body). The mean Hounsfield unit (HU) number within the medullary part of the vertebral bodies of L1 and L2 is used to estimate BMD. From reference [32] with permission
to apply it to more routine clinical CT images such as CT colonography [35–37]. This is an exciting prospect, but computation times currently limit its current application in clinical practice. Its potential use in clinical practice will become more practical as computational power inevitably increases. The notion of opportunistic screening for other diseases at CT is neither a new concept nor limited to osteoporosis. L1 attenuation has also been studied along with radiographic sarcopenia metrics (such as psoas major muscle volume) in the setting of prostate cancer to predict non-cancer-related death and 10-year fracture probability [38, 39]. CT measurements of visceral fat have also been linked to the presence of metabolic disease risk factors [40, 41]. Associations between coronary artery disease, type II diabetes, and abdominal calcification have been observed as well [40, 42, 43]. Despite this evidence, research on the ability of abdominal CT-derived measurements such as abdominal aortic calcification and visceral fat volumes to predict future cardiovascular event risk is lacking. If quantitative disease markers for colorectal cancer, osteoporosis, and cardiovascular disease could be obtained from CT colonography, it may be possible to combine this
information with clinical history and laboratory data to generate individualized predictions of fracture and cardiovascular event risks. Such predictions would have to go through rigorous cost-benefit analysis and validation before being applied routinely in clinical practice, and more research is needed in this area. Furthermore, as with many radiologic exams, the risk of over-diagnosis would still be present.
Conclusion In summary, opportunistic osteoporosis screening using L1 trabecular attenuation has been studied in several independent populations and can reliably identify patients who would likely have osteoporosis based on current clinical standard of DXA. Although not conclusively proven, CT-derived L1 attenuation may actually be as good or better as DXA for identifying patients at greatest risk for future fracture. However, because technical variations in spinal attenuation measurements currently exist, L1 attenuation should not be used in place of DXA or qCT at this time. Rather, patients with
Clinic Rev Bone Miner Metab
severely low L1 attenuation (such as ≤ 90 HU) should be strongly urged to pursue further BMD workup to address and potentially reduce fracture risk. We have successfully implemented this practice at our institution and have received positive feedback from referring providers due to identification of previously unscreened patients with severe fracture risk. Asynchronous QCT (CTXA) represents an FDA-approved method for deriving a DXA-equivalent femoral neck BMD and T-score from CT scans performed for other indications. These measures can be obtained retrospectively or prospectively for patients undergoing routine abdominal CT. Finally, as technologies such as finite element analysis, automated trabecular and muscle segmentation algorithms, and abdominal aortic calcium scoring become more developed, routine body CT stands to become a rich source of previously unused, valuable data that can stratify risk and support individuals and clinicians in making personalized healthcare decisions.
9.
10.
11.
12.
13.
14.
15.
16. Compliance with Ethical Standards Funding This study supported in part by the Institute for Clinical and Translation Research (ICTR) at the University of Wisconsin—Madison.
17.
Conflict of Interest Dr. Pickhardt is co-founder of VirtuoCTC; advisor to Bracco and Check-Cap; and shareholder in SHINE, Elucent, and Cellectar Biosciences. The authors declare no relevant conflicts of interest for this work.
18.
Ethical Approval and Informed Consent N/A (review article).
19.
References
20.
1. 2. 3.
4.
5.
6.
7.
8.
Kanis JA, Borgstrom F, De Laet C, et al. Assessment of fracture risk. Osteoporos Int. 2005;16:581–9. Cauley JA, Thompson DE, Ensrud KC, et al. Risk of mortality following clinical fractures. Osteoporos Int. 2000;11:556–61. U.S. Department of Health and Human Services. Bone health and osteoporosis: a report of the Surgeon General. Rockville: U.S. Department of Health and Human Services, Office of the Surgeon General; 2004. Wang Z, Ward MM, Chan L, Bhattacharyya T. Adherence to oral bisphosphonates and the risk of subtrochanteric and femoral shaft fractures among female medicare beneficiaries. Osteoporos Int. 2014;25:2109–16. Curtis JR, Carbone L, Cheng H, et al. Longitudinal trends in use of bone mass measurement among older americans, 1999-2005. J Bone Miner Res. 2008;23:1061–7. Cohen K, Maier D. Osteoporosis: evaluation of screening patterns in a primary-care group practice. J Clin Densitom. 2008;11:498– 502. Mettler FA, Thomadsen BR, Bhargavan M, et al. Medical radiation exposure in the U.S. in 2006: preliminary results. Health Phys. 2008;95:502–7. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data. Radiology. 2016;278:563–77.
21.
22.
23.
24.
25.
26. 27.
Pickhardt PJ, Pooler BD, Lauder T, et al. Opportunistic screening for osteoporosis using abdominal computed tomography scans obtained for other indications. Ann Intern Med. 2013;158:588–95. Steiger P, Block JE, Steiger S, et al. Spinal bone mineral density measured with quantitiative CT - effect of region of interest, vertebral level, and technique. Radiology. 1990;175:537–43. Lee SJ, Binkley N, Lubner MG, et al. Opportunistic screening for osteoporosis using the sagittal reconstruction from routine abdominal CT for combined assessment of vertebral fractures and density. Osteoporos Int. 2016;27:1131–6. Carberry GA, Pooler BD, Binkley N, et al. Unreported vertebral body compression fractures at abdominal multidetector CT. Radiology. 2013;268:120–6. Pickhardt PJ, Lee SJ, Muñoz Del Rio A, et al. Simultaneous screening for osteoporosis at CT colonography: bone mineral density assessment using MDCT attenuation techniques compared with the DXA reference standard. J Bone Miner Res. 2011;26:2194–203. Buckens CF, Dijkhuis G, de Keizer B, et al. Opportunistic screening for osteoporosis on routine computed tomography? An external validation study. Eur Radiol. 2015;25:2074–9. Alacreu E, Moratal D, Arana E. Opportunistic screening for osteoporosis by routine CT in southern Europe. Osteoporos Int. 2017;28: 983–90. Burke CJ, Didolkar MM, Barnhart HX, Vinson EN. The use of routine non-density calibrated clinical computed tomography data as a potentially useful screening tool for identifying patients with osteoporosis. Clin Cases Miner Bone Metab. 2016;13(2):135–40. Majumdar SR, Leslie WD. Conventional computed tomography imaging and bone mineral density: opportunistic screening or Bincidentaloporosis^? Ann Intern Med. 2013;158:630–1. Graffy PM, Lee SJ, Ziemlewicz TJ, Pickhardt PJ. Prevalence of vertebral compression fractures on routine body CT scans according to L1 trabecular attenuation: determining relevant thresholds for opportunistic osteoporosis screening. Am J Roentgenol. 2017;209:1–6. Lee SJ, Anderson PA, Pickhardt PJ. Predicting future hip fractures at routine abdominal CT using opportunistic osteoporosis screening measures: a matched case-control study. Am J Roentgenol. 2017;209:1–8. Pompe E, de Jong PA, de Jong WU, et al. Inter-observer and interexamination variability of manual vertebral bone attenuation measurements on computed tomography. Eur Radiol. 2016;26:3046– 53. Gerety E-L, Hopper MA, PWP B. The reliability of measuring the density of the L1 vertebral body on CT imaging as a predictor of bone mineral density. Clin Radiol. 2017;72:177.e9–177.e15. Pompe E, Willemink MJ, Dijkhuis GR, et al. Intravenous contrast injection significantly affects bone mineral density measured on CT. Eur Radiol. 2015;25:283–9. Pickhardt PJ, Lauder T, Pooler BD, et al. Effect of IV contrast on lumbar trabecular attenuation at routine abdominal CT: correlation with DXA and implications for opportunistic osteoporosis screening. Osteoporos Int. 2016;27:147–52. Garner HW, Paturzo MM, Gaudier G, et al. Variation in attenuation in L1 trabecular bone at different tube voltages: caution is warranted when screening for osteoporosis with the use of opportunistic CT. Am J Roentgenol. 2017;208:165–70. Lamba R, McGahan JP, Corwin MT, et al. CT Hounsfield numbers of soft tissues on unenhanced abdominal CT scans: variability between two different manufacturers’ MDCT scanners. Am J Roentgenol. 2014;203:1013–20. Adams JE. Quantitative computed tomography. Eur J Radiol. 2009;71:415–24. Riggs BL, Melton LJ, Robb RA, et al. A population-based assessment of rates of bone loss at multiple skeletal sites: evidence for
Clinic Rev Bone Miner Metab substantial trabecular bone loss in young adult women and men. J Bone Miner Res. 2007;23:205–14. 28. Brown JK, Timm W, Bodeen G, et al. Asynchronously calibrated quantitative bone densitometry. J Clin Densitom. 2017;20:216–25. 29. Summers RM, Baecher N, Yao J, et al. Feasibility of simultaneous computed tomographic colonography and fully automated bone mineral densitometry in a single examination. J Comput Assist Tomogr. 2011;35:212–6. 30. Pickhardt PJ, Bodeen G, Brett A, et al. Comparison of femoral neck BMD evaluation obtained using lunar DXA and QCT with asynchronous calibration from CT Colonography. J Clin Densitom. 2015;18:5–12. 31. Ziemlewicz TJ, Maciejewski A, Binkley N, et al. Opportunistic quantitative CT bone mineral density measurement at the proximal femur using routine contrast-enhanced scans: direct comparison with DXA in 355 adults. J Bone Miner Res. 2016;31:1835–40. 32. Ziemlewicz TJ, Maciejewski A, Binkley N, et al. Direct comparison of unenhanced and contrast-enhanced CT for opportunistic proximal femur bone mineral density measurement: implications for osteoporosis screening. Am J Roentgenol. 2016;206:694–8. 33. Ziemlewicz TJ, Binkley N, Pickhardt PJ. Opportunistic osteoporosis screening: addition of quantitative CT bone mineral density evaluation to CT Colonography. J Am Coll Radiol. 2015;12: 1036–41. 34. Pompe E, Bartstra J, Verhaar HJ, et al. Bone density loss on computed tomography at 3-year follow-up in current compared to former male smokers. Eur J Radiol. 2017;89:177–81. 35. Fidler JL, Murthy NS, Khosla S, et al. Comprehensive assessment of osteoporosis and bone fragility with CT Colonography. Radiology. 2016;278:172–80.
36.
Keaveny TM. Biomechanical computed tomography-noninvasive bone strength analysis using clinical computed tomography scans. Ann N Y Acad Sci. 2010;1192:57–65. 37. Weber NK, Fidler JL, Keaveny TM, et al. Validation of a CTderived method for osteoporosis screening in IBD patients undergoing contrast-enhanced CT enterography. Am J Gastroenterol. 2014;109:401–8. 38. McDonald AM, Swain TA, Mayhew DL, et al. CT measures of bone mineral density and muscle mass can be used to predict noncancer death in men with prostate cancer. Radiology. 2017;282:475–83. 39. McDonald AM, Jones JA, Cardan RA, et al. Combining computed tomography-based bone density assessment with FRAX screening in men with prostate cancer. J Clin Densitom. 2016;19:430–5. 40. Fox CS, Massaro JM, Hoffmann U, et al. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham heart study. Circulation. 2007;116:39–48. 41. Pickhardt PJ, Jee Y, O’Connor SD, del Rio AM. Visceral adiposity and hepatic steatosis at abdominal CT: association with the metabolic syndrome. AJR Am J Roentgenol. 2012;198:1100–7. 42. Wong ND, Lopez VA, Allison M, et al. Abdominal aortic calcium and multi-site atherosclerosis: the multiethnic study of atherosclerosis. Atherosclerosis. 2011;214:436–41. 43. Reaven PD, Sacks J. Coronary artery and abdominal aortic calcification are associated with cardiovascular disease in type 2 diabetes. Diabetologia. 2005;48:379–85.