Abstract
Objectives:
The objective of this work was to evaluate the incremental value of MR angiography over plain radiographs and MRI for the differentiation of aneurysmal bone cysts (ABCs) from unicameral bone cysts (UBCs).
Methods:
Thirty-six juvenile patients with histologically secured primary ABCs or UBCs were included in this retrospective study. Two radiologists assessed all obtained images in a blinded fashion using a catalog of previously suggested imaging findings. A second readout with supplementary MR angiography images was performed after 8 weeks to prevent observer recall bias. Diagnostic accuracy parameters were calculated for individual imaging findings, and overall diagnostic accuracy and diagnostic confidence were assessed for all readouts. Receiver operating characteristic (ROC) curve comparison was used to determine the incremental value of MR angiography.
Results:
Of 16 imaging features, only abnormal vascularization in MR angiography provided sufficient diagnostic accuracy for the identification of ABCs. Other imaging features such as fluid–fluid levels and internal septations were insufficient for the differentiation of UBCs from ABCs. Availability of MR angiography images significantly increased diagnostic accuracy (94.4 vs 75.0% and 83.3 vs 69.4%, respectively, p < 0.05) and diagnostic confidence (4.5 vs 3.7, p < 0.05) of reading radiologists.
Conclusion:
The presence of arterial feeders in MR angiography can accurately discriminate primary ABCs from UBCs and increases the diagnostic accuracy and diagnostic confidence of reporting radiologists.
Advances in knowledge:
Radiographic differentiation of cystic bone lesions such as ABCs and UBCs remains challenging. We demonstrate that MR angiography provides incremental value and suggest inclusion in standard examination protocols.
Introduction
Aneurysmal bone cysts (ABC) were described by Jaffé and Lichtenstein more than 70 years ago as benign, tumor-like bone lesions consisting of blood-filled cavities bounded by thin septa. 1–3 Their name is based solely on the initial histopathological morphology, and the pathological mechanism of their formation has not been conclusively elucidated to date. Among other theories, USP6-driven misguided bone maturation processes, circulatory disturbances, and sequelae of trauma are discussed. 4–8
ABCs usually present within the first 20 years of life with swelling, pain, pathologic fractures, or as an incidental finding, typically in the long tubular bones, pelvis, or spine. 9–11 Due to their locally aggressive growth, they require histopathological confirmation to exclude malignancy. Therefore, the therapeutic gold-standard for the treatment of ABCs remains surgical excision. Unicameral bone cysts (UBCs) do not require histologic confirmation and surgical excision is not always necessary. 12,13 The differentiation of ABCs from UBCs remains challenging due to similar clinical symptoms and overlap in location and appearance. The high prevalence of pathological fractures in cystic bone lesions that cannot always be distinguished from the erosion of cortical bone further complicates distinction.
Radiographically, ABCs have been described as lobulated cystic lesions with eccentric growth, distension of the bone, and a thin or destroyed cortical border. MRI studies also show internal septa and occasional fluid–fluid levels within the cystic spaces. After applying contrast medium, enhancement of the cyst walls and the septa can be observed. The signal behavior varies within the cystic component, depending on protein content and accumulation of cellular debris, but is generally high on T 2- and low on T 1 weighted images. Sometimes, a sclerotic rim around the ABC is present. UBCs have been described as centrally located and well-marginated lesions with homogenous signal intensities and without the presence of septations or fluid–fluid levels. 14–19 Even with multiple imaging modalities available, a sensitivity of around 80% and a specificity of around 70% for the differentiation of ABCs from other cystic bone lesions has been reported. 2,14,20,21 Evaluating novel methods to differentiate UBCs more accurately from ABCs can therefore make an important contribution to sparing the predominantly young patient population with UBCs unnecessary surgery and prolonged immobilization.
Therefore, the purpose of this study was to evaluate the diagnostic accuracy of a catalog of imaging findings and the incremental value of MR angiography for the differentiation of UBCs from primary ABCs.
Methods and materials
The institutional review board approved this retrospective study. The requirement to obtain written informed consent was waived. When pictures were used, written consent was obtained.
Patient selection
Patients under the age of 25 who presented with cystic bone lesions between January of 2014 and December 2021 were considered eligible for study inclusion. Medical records of all eligible patients were assessed for surgical treatment and histological confirmation of the suspected lesions. Exclusion criteria were age over 25 years, insufficient image quality, and bone lesions different from primary ABCs and UBCs. Detailed information is given in Figure 1 .
Figure 1.
STARD (Standards for Reporting of Diagnostic Accuracy Studies) flow chart of patient inclusion.
Aneurysmal bone cysts
ABCs were considered primary if they carried an USP6 rearrangement and did not show histological features of other bone lesions. ABCs were considered secondary if they occurred on basis of other bone lesions of following trauma with a time gap <1 year. No discrimination was made between classical ABCs and solid variant ABCs.
Imaging protocol
For all patients, plain-film radiographs were obtained. 30 patients underwent additional MRI to assist pre-operative planning. In six patients, MRI was not performed due to patient refusal. MRI was conducted on different systems of the same manufacturer (Siemens Healthcare, Erlangen, Germany): AvantoFit (1.5 Tesla, 04/2013–today), Trio (3 Tesla, 01/2008–10/2013) or PrismaFit (3 Tesla, 11/2013–today). Examinations included pre- and, in 27 cases, post-contrast axial T 1 weighted turbo spin-echo sequences with fat suppression and proton density-weighted sequences with fat suppression, all in the transversal, sagittal and coronal planes (slice thickness: 4 mm). Pulse sequence parameters (echo time, repetition time, flip angle, etc.), field of view, and acquisition matrix were adapted for every examination and body region and varied between the scanner types. Before surgical resection, TWIST-MR angiography was performed for the assessment of feeder-arteries and to assess the requirement for angiographic devascularization, (TR: 3.31 ms, TE: 1.3 ms, Flip angle: 24°, ST: 0.8 mm). Injection of macrocyclic contrast agents Gadobutrol (Gadovist, Bayer Healthcare, Leverkusen, Germany) or Gadoterate (Dotarem, Guerbet, Villepinte, France) adapted to the patient’s weight was performed using a power injector (Accutron MR, Medtron, Saarbruecken, Germany) at a rate of 2 ml s−1, followed by the application of 20 ml saline at a rate of 2 ml s−1.
Image interpretation
Two radiologists (K.E. and L.G.) with 14 and 3 years of experience in musculoskeletal imaging analyzed all images blinded to patient identities and histologic results using a dedicated catalog of 16 typical findings in the setting of ABCs and UBCs. 14,15,18 MR images were primarily used for image assessment; however, if not available, plain-film radiographs were used. Two readouts were independently performed by both radiologists with a time interval of 8 weeks to prevent observer recall bias. The images were presented in random order for both readouts. The first readout was performed without MR angiography image sets, which were subsequently added in the second readout. In case of disparate readings, a third radiologist with 8 years of experience in musculoskeletal imaging was consulted. Reported is the majority reading.
Cyst signal intensities were compared with adjacent muscles by manually placing a region of interest (ROI) in the cysts and adjacent muscle tissue. Isointense signal was defined as a deviation of the intensity of no more than 20% from muscle signal, whereas hyper- and hypointense signal was defined as a deviation of more than 20%.
Statistical analysis
Statistical analysis was performed with dedicated commercial software (Prism 9 for macOS, V. 9.0.1, GraphPad Software LLC, San Diego; MedCalc for Windows, V. 20.022, MedCalc, Mariakerke, Belgium). Differences in baseline characteristics were assessed using t-tests, if applicable, or χ2 tests. Differences in paired variables were assessed using Wilcoxon matched-pairs signed-rank test. Interrater agreement was calculated using Cohens κ.
Imaging findings were dichotomized and compiled in cross-tables. Sensitivity, specificity, positive-predictive value (PPV), negative-predictive value (NPV), and odds ratio for the detection of ABCs were calculated for all obtained imaging features. Receiver operating characteristic (ROC) curve analysis was used to evaluate associations of sensitivity, specificity, and predictive values with the amount of imaging features present. For both readers and readouts, sensitivity, specificity, PPV, NPV, accuracy and AUC were calculated. ROC curve comparison was used to determine the incremental value of MR angiography for both readers.
Results
Patient demographics
Of 138 patients who underwent imaging for suspected ABCs or UBCs, 101 patients were excluded due to age (n = 62), insufficient image quality (n = 2), or lesions different from ABCs and UBCs (n = 38). The final study population consisted of 36 patients with a mean age of 13.6 years (range, 2–25 years) (Figure 1). The most common site for the presence of ABCs and UBCs was the proximal humerus (52.6% and 47.1%, respectively), followed by the fibula for ABCs (21.1%) and the femur (23.5%) for UBCs. Eight patients with ABCs (42.1 %) and 10 patients with UBCs (58.8 %) had a fracture at the lesion site on initial presentation. Two patients with primary ABCs sustained trauma at the same site with a time gap exceeding 1 year. No statistical significance was observed between the demographics of both groups. Detailed patient characteristics are depicted in Table 1.
Table 1.
Detailed patient characteristics
| Variables- n (%) or mean ± SD | Aneurysmal bone cyst (n = 19) | Unicameral bone cyst (n = 17) | p-value |
|---|---|---|---|
| Age (years) | 12.21 ± 7.9 | 15.4 ± 6.4 | 0.20 |
| Sex (n) | |||
|
5 (26.3 %) | 4 (23.5 %) | 0.93 |
|
14 (73.7 %) | 13 (76.5 %) | 0.87 |
| Site of lesion (n) | |||
|
10 (52.6 %) | 8 (47.1 %) | 0.82 |
|
1 (5.3 %) | 1 (5.9 %) | 0.99 |
|
1 (5.3 %) | 1 (5.9 %) | 0.99 |
|
1 (5.3 %) | 4 (23.5%) | 0.72 |
|
1 (5.3 %) | 1 (5.9 %) | 0.99 |
|
4 (21.1 %) | 1 (5.9 %) | 0.75 |
|
1 (5.3 %) | 1 (5.9 %) | 0.99 |
| Fracture on presentation | 8 (42.1 %) | 10 (58.8 %) | 0.49 |
SD, standard deviation.
No statistical difference was observed between patient age, gender, distribution of the lesions, or fractures on initial presentation between both groups.
Imaging features
Imaging features showed substantial overlap between UBCs and ABCs, indicating that most individual image features cannot reliably differentiate ABCs and UBCs. Often cited features such as bone expansion, internal septation, and fluid–fluid levels showed a high sensitivity for detecting ABCs, but overall specificity was low for the differentiation from UBCs, and overlap was high. The highest specificity of features for the differentiation of ABCs from UBCs was the presence of a periosteal reaction and the presence of an extraosseous soft tissue component (Table 2, Figure 2), however, statistical significance was not reached due to strong overlap. For individual imaging features, no significant differences were observed between the sites of the lesion. ROC curve analysis demonstrated an increase in specificity depending on the number of present imaging features, with 50.0% specificity when more than 10 features were present and 83.3% specificity when more than 12 features were present (Figure 3). Cyst signal in T 1 weighted and PD-weighted images also showed similar intensities for ABCs and UBCs, with T 1 weighted images being mostly isointense and PD-weighted images being mostly hyperintense to muscle tissue. Agreement between readers for the identification of individual imaging features was high with a weighted κ of 0.71 (CI, 0.48–0.95). Detailed imaging features are listed in Tables 2 and 3.
Table 2.
Diagnostic accuracy parameters for the differentiation of ABCs from UBCs
| Imaging features (%) — (95% confidence interval) |
Sensitivity | Specificity | PPV | NPV | Odds ratio | p-value |
|---|---|---|---|---|---|---|
| Bone expansion | 94.7 (75.4–99.7) |
6.3 (0.3–28) |
54.6 (38.0–70.2) |
50.0 (2.6–97.4) |
1.20 (0.06–23.9) |
> 0.99 |
| Circumscribed lesion | 36.8 (19.2–59.0) |
17.7 (6.2–41.0) |
33.3 (17.2–54.6) |
20.0 (7.1–45.2) |
0.13 (0.03–0.60) |
0.01 |
| Internal septa | 79.0 (56.7–91.5) |
18.8 (6.6–43.0) |
53.6 (35.8–70.5) |
42.9 (15.8–75.0) |
0.87 (0.19–3.74) |
> 0.99 |
| Peripheral lobulation | 83.3 (60.1–94.2) |
31.3 (14.2–55.6) |
57.7 (39.0–74.5) |
83.3 (30.6–86.3) |
2.30 (0.42–9.80) |
0.43 |
| Lytic growth | 94.7 (74.5–99.7) |
11.8 (2.1–34.3) |
54.6 (38.0–70.2) |
66.7 (11.9–98.3) |
2.40 (0.25–35.5) |
0.59 |
| Periostal reaction | 21.1 (8.5–43.3) |
87.5 (64.0–97.8) |
66.7 (30.0–94.1) |
48.3 (31.4–65.6) |
1.87 (0.36–10.82) |
0.67 |
| Thinned or damaged cortical bone | 94.7 (75.4–99.7) |
18.8 (6.6–43.0) |
58.1 (40.8–73.6) |
75.0 (30.0–98.7) |
4.15 (0.54–56.56) |
0.31 |
| Extraosseous soft tissue component | 11.8 (2.1–34.3) |
100 (75.6–100) |
100.0 (17.8–100) |
44.4 (27.6–62.7) |
∞ (0.33 – ∞) |
0.50 |
| Eccentric location | 89.5 (68.6–98.1) |
6.7 (0.3–30.0) |
54.8 (37.8–70.8) |
33.3 (17.1–88.2) |
0.61 (0.04–5.72) |
> 0.99 |
| Hypointense rim (T1) | 93.8 (71.7–99.7) |
10.0 (0.5–40.4) |
62.5 (42.7–78.8) |
50.0 (2.6–97.4) |
1.67 (0.08–33.51) |
> 0.99 |
| Perilesional edema | 52.9 (30.7–73.8) |
63.6 (35.4–84.8) |
69.2 (42.4–87.3) |
46.7 (24.8–69.9) |
1.97 (0.44–7.66) |
0.46 |
| Solid component | 58.8 (36.0–78.4) |
72.7 (43.4–90.3) |
76.9 (49.7–91.8) |
53.3 (30.1–75.2) |
3.81 (0.80–16.2) |
0.14 |
| Fluid–fluid level | 70.6 (46.9–86.7) |
54.6 (28.0–78.7) |
70.6 (47.9–86.7) |
54.6 (28.0–78.3) |
2.88 (0.60–11.71) |
0.25 |
| Cyst wall enhancement | 93.8 (71.7–99.7) |
11.1 (0.6–43.5) |
65.2 (44.9–81.2) |
50.0 (2.6–97.4) |
1.88 (0.09–37.68) |
> 0.99 |
| Internal septal enhancement | 92.9 (68.5–99.6) |
11.1 (0.6–43.5) |
61.9 (40.2–79.3) |
50.0 (3.6–97.4) |
1.63 (0.08–32.96) |
> 0.99 |
| Presence of arterial feeder |
72.7
(43.4–90.3) |
88.9
(56.5–99.4) |
88.9
(56.5–99.4) |
72.7
(43.4–90.3) |
21.33
(2.15–25.9) |
0.01 |
With the exception of the presence of feeding arteries, no imaging feature provided sufficient specificity or NPV to rule out ABCs. No significant differences were observed for the occurrence of the imaging features between ABCs and UBCs. Corresponding 95% confidence intervals are given in brackets. ABC, aneurysmal bone cyst; UBC, unicameral bone cyst; NPV, negative-predictive value; PPV, positive-predictive value. T 1WI, T 1 weighted image
Figure 2.
PD-weighted images of histopathological secured unicameral (a, b) and aneurysmal (c, d) bone cysts. Note the overlap of imaging features usually attributed to ABCs such as fluid–fluid levels (arrows), lobulation, and septation (arrowheads). A small soft-tissue component is present at the inferior border of the UBC (asterisk). ABC, aneurysmal bone cyst; PD, proton density; UBC, unicameral bone cyst.
Figure 3.
For the differentiation of ABCs from UBCs, ROC curve analysis demonstrates an increase in specificity at cost of sensitivity depending on the number of imaging features present. The marking in the lower left corner marks specificity = 100%, sensitivity = 0%; the marking in the top right corner marks specificity = 0%, sensitivity = 100%. AUC was 0.65 (p = 0.15). ABC, aneurysmal bone cyst; AUC, area under the curve; ROC, receiver operator characteristic; UBC, unicameral bone cyst.
Table 3.
MRI cyst signal intensities of ABCs and UBCs
| ABC | UBC | ||||||
|---|---|---|---|---|---|---|---|
| Cyst signal (%) | Hypointense | Isointense | Hyperintense | Hypointense | Isointense | Hyperintense | p-value |
| Total | |||||||
|
15.8% | 68.4% | 15.8% | 5.9% | 82.4% | 11.7% | 0.57 |
|
10.5% | 0.0% | 89.5% | 0.0% | 0.0% | 100.0% | 0.18 |
| With fracture | |||||||
|
12.5% | 50.0% | 37.5% | 0.0% | 80.0% | 20.0% | 0.34 |
|
12.5% | 0.0% | 87.5% | 0.0% | 0.0% | 100.0% | 0.26 |
| W/o fracture | |||||||
|
18.2% | 81.8% | 0.0% | 14.3% | 85.7% | 0.0% | 0.84 |
|
9.1% | 0.0% | 90.9% | 0.0% | 0.0% | 100.0% | 0.43 |
ABC, aneurysmal bone cyst; PD, proton density;T 1W, T 1 weighted; UBC, unicameral bone cyst.
No significant differences were observed between ABCs and UBCs regarding cyst signal intensity.
MR angiography and presence of feeding arteries
TWIST-MR angiography was routinely obtained to display vascular supply and identify feeding arteries for assessing the requirement for preoperative embolization. Notably, ABCs showed on average 1.4 (range, 0–3) abnormal feeding arteries arising from adjacent parent arteries, whereas only one UBC showed increased vascular supply via a feeding artery compared to the adjacent bone (range, 0–1) (Figures 4 and 5). Consequently, the presence of feeding arteries provided high sensitivity, specificity, PPV, and NVP for the differentiation of ABCs from UBCs (Table 2).
Figure 4.
PD-weighted images of unicameral (a, b, c) and aneurysmal (d, e, f) bone cysts. Again, there is a substantial overlap of imaging features usually attributed to ABCs, with septation, lobulation, and fluid-fluid levels (arrows) present in the UBC. However, MR angiography reveals normal vascularity of the UBC (c), whereas vascularity is markedly increased in the ABC (asterisk) (f). This patient went on to have pre-operative embolization followed by complete surgical excision of the ABC. ABC, aneurysmal bone cyst; PD, proton density; UBC, unicameral bone cyst.
Figure 5.
PD-weighted images of an aneurysmal bone cyst with typical features (a). MR angiography and conventional angiography reveal an increase in vascularization with countless arterial feeders (circle) arising from the anterior humeral circumflex artery and the deltoid branch of the axillary artery (arrows) (b, c). Embolization was performed prior to surgical excision (asterisks) (d). Complete devascularization was confirmed intraoperatively (e). PD, proton density.
Diagnostic accuracy
Diagnostic parameters of both readers for the differentiation of ABCs from UBCs are shown in Table 4. The availability of MR angiography images provided significant incremental value, with increases in sensitivity, specificity, PPV, and NPV for both readers (p < 0.005 and p = 0.019, respectively) (Figure 6, Table 4). Furthermore, the diagnostic confidence was rated significantly higher by all readers when imaging protocols included MR angiography (p < 0.005) (Table 4). Interreader agreement for the identification of ABCs and UBCs increased from a weighted κ of 0.63 (CI, 0.49–0.74) without MR angiography to a weighted κ of 0.79 (CI, 0.66–0.90) with MR angiography.
Table 4.
Diagnostic performance indices for both readers with and without MR angiography
| Accuracy parameters (%) — (95% confidence interval) | Sensitivity | Specificity | PPV | NPV | Accuracy | AUC | p-value (AUC) | Diagnostic confidence | p-value (DC) |
|---|---|---|---|---|---|---|---|---|---|
| Reader 1 | |||||||||
|
75.0 (50.9–91.3) |
75.0 (47.6–92.7) |
78.9 (60.7–90.1) |
70.6 (51.6–84.3) |
75.0 (57.8–87.9) |
0.75 (0.58–0.88) |
<0.005 | 3.7 (3.4–4.1) |
<0.005 |
|
95.0 (75.1–99.9) |
93.8 (69.8–99.8) |
95.0 (73.9–99.2) |
93.8 (68.9–99.0) |
94.4 (81.3–99.3) |
0.94 (0.81–0.99) |
<0.005 | 4.5 (4.2–4.8) |
<0.005 |
| Reader 2 | |||||||||
|
70.0 (45.7–88.1) |
68.8 (41.3–89.0) |
73.7 (56.2–86.0) |
64.7 (46.5–79.5) |
69.4 (51.9–83.7) |
0.69 (0.52–0.84) |
0.019 | 3.6 (3.3–3.9) |
<0.005 |
|
84.2 (60.4–96.2) |
82.4 (56.6–96.2) |
84.2 (65.2–93.8) |
82.4 (61.8–93.1) |
83.3 (67.2–93.3) |
0.83 (0.67–0.94) |
0.019 | 4.5 (4.2–4.7) |
<0.005 |
AUC, area under the curve; DC, diagnostic confidence; NPV, negative-predictive value; PPV, positive-predictive value.
Sensitivity, specificity, PPV, NPV, accuracy, AUC, and diagnostic confidence increased significantly for both readers when supplementary MR angiography was available. Corresponding 95% confidence intervals are given in brackets.
Figure 6.
Results of ROC curve analysis for both readers without (dotted line) and with (solid line) MR angiography imaging. Availability of MR angiography for readouts significantly increased the AUC of both reporting radiologists (p = 0.005 and 0.019, respectively). AUC, area under the curve; ROC, receiver operator characteristic.
Discussion
This study demonstrates the added value that MR angiography provides for the diagnosis of primary ABCs with emphasis on the differentiation from UBCs. We found that the presence of feeding arteries in MR angiography can accurately discriminate ABCs from UBCs and that the availability of MR angiography images increases both the diagnostic accuracy and the diagnostic confidence of reporting radiologists.
UBCs are an important differential diagnosis in assessing cystic and expansive growing bone lesions. They manifest with similar symptoms and in similar demographic groups compared to ABCs. No significant differences were observed in our study population regarding patient age, gender, lesion localization, or the percentage of fractures on presentation. In contrast to ABCs or malignant bone tumors with cystic components, however, UBCs do not grow locally destructive, and expansion is generally slow. Therefore, watch and wait is an acceptable therapeutic approach when treating UBCs, whereas early treatment should be sought for suspected ABCs to obtain tissue samples and prevent locally destructive growth. 22 The capability to accurately differentiate ABCs and UBCs without surgical intervention is therefore highly relevant and can spare the predominantly young patient collective unnecessary surgery and immobilization.
To assess cystic bone lesions, a combination of plain film radiographs and MR imaging is considered standard. However, MR angiography is routinely performed at large centers before surgical resection to display feeding arterial vessels and assess whether angiographic devascularization is required. 12,13 Certain imaging features, such as internal septa within the cyst and fluid–fluid levels have previously been described as characteristic for ABCs and are still considered central diagnostic criteria. 15 However, it has since been suggested that they occur in various bone and soft tissue lesions and are therefore not specific for the identification of ABCs. 23 In our study, we demonstrate high overlap between almost all imaging findings that have previously been considered characteristic for ABCs, both in plain film radiographs and MRI. Consequently, individual imaging features lack sensitivity and specificity and should not be relied upon for the differentiation of ABCs from UBCs, although an increase in the amount of total imaging features present increases specificity as demonstrated by ROC-curve analysis. In this context, it is important to highlight the distortion of imaging features resulting from fractures, which are the main reason patients with UBCs become symptomatic and seek treatment. Fractures can mimic permeative changes of cortical bone, cause intralesional bleeding that creates fluid–fluid levels and alter MR signal intensities, all of which complicate the differentiation of UBCs from ABCs and other cystic bone lesions. In our study, 58.8% of patients with UBCs presented with an acute fracture, compared to 63–87% previously reported. 24 UBCs in patients without fracture were primarily incidental findings. The number of patients with ABCs that presented with an acute fracture was slightly lower, mainly because patients sought treatment before a pathological fracture occurred due to symptoms that can be attributed to the expansive growth of ABCs such as swelling and pain.
Previous studies reported a sensitivity of 82.6% and a specificity of 70% for identifying ABCs using plain film radiographs and MRI. 15 Albeit not directly comparable due to the restriction of this study on ABCs and UBCs, we reached a comparable diagnostic accuracy with 75.0% and 69.4% before MR angiography. However, diagnostic confidence of both raters was only moderate with values of 3.7 and 3.6, respectively, which underlines how difficult the differentiation of ABCs and UBCs remains for radiologists, especially in the setting of pathological fractures. Notably, both readers' diagnostic accuracy and diagnostic confidence increased significantly when MR angiography was available for readouts, reaching a diagnostic accuracy of 94.4% and 83.3 %, respectively, as well as a diagnostic confidence of 4.5 for both readers. In MR angiography, most ABCs showed hypervascularization resulting from 1 to 3 abnormal feeding arteries arising from adjacent parent arteries. UBCs, in contrast, did not show increased vascular supply or feeding arteries compared to the adjacent bone with the exception of one patient, in which a feeding artery could not be excluded angiographically. This finding seems plausible, considering that primary ABCs are considered a vascular neoplasm primarily. 1 A separate analysis confirmed that the presence of arterial feeding arteries was the only imaging criterion that provided significant diagnostic accuracy for the differentiation of UBCs from ABCs with a specificity of 88.9%.
Due to high recurrence rates of curettage, surgical treatment of ABCs is increasingly performed via en-bloc resection and bone interposition, preceded by pre-operative embolization to reduce bleeding. 22,25–27 Prior MR angiography can increase the success rate of angiographic devascularization and reduce the duration of the intervention. To increase diagnostic accuracy and aid pre-interventional and pre-operative planning, we therefore strongly suggest to routinely perform MRI with MR angiography to evaluate the presence of arterial feeders when assessing cystic bone lesions.
In this context, we want to highlight that other cystic bone lesions can appear different on plain film radiographs and MRI, and that the presence of feeding arteries should not be considered specific for primary ABCs but rather a criterion that makes the presence of UBCs unlikely. Therefore, when feeding arteries are present, tissue samples for the histological examination should always be obtained to exclude potentially malignant tumors such as telangiectatic osteosarcoma, which also demonstrate increased vascularization and share many imaging features with ABCs. 28
This study has certain limitations we would like to address. First, with only 36 patients, our study population is relatively small, and statistical significance was not always reached. While we believe this is mainly due to overlap in imaging features, we cannot exclude the possibility that the differences will become more apparent in larger patient cohorts and reach statistical significance. Second, we only included patients with primary or true ABCs. Since ABCs can also arise as sequelae from other bone lesions, the vascularization pattern of secondary ABCs could be different. Third, a high percentage of our study population presented with pathological fractures, which can distort imaging findings. However, since pathological fractures are the main reason cystic bone lesions become symptomatic, our findings are likely comparable to the imaging findings in daily practice. Finally, despite a time interval of 8 weeks between readouts and random order of the presented images, we cannot exclude some extent of observer recall bias in our second readout.
In conclusion, our study demonstrates the incremental value of MR angiography over plain film radiographs and conventional MRI for both the diagnosis and treatment of primary ABCs, with an emphasis on the differentiation from UBCs. We demonstrate that most imaging findings previously considered characteristic for ABCs lack sensitivity and specificity and should not be relied upon for the differentiation of ABCs from UBCs, especially in the setting of pathological fractures. In contrast, the presence of arterial feeders in MR angiography can accurately discriminate ABCs from UBCs and increase diagnostic accuracy and diagnostic confidence of reporting radiologists. Their depiction should therefore be included in MRI protocols in the future.
Footnotes
Conflict of Interest: C.B. and I.Y. received speaking fees from Siemens Healthineers.
The other authors have no conflict of interest to disclose.
Funding: Open Access funding enabled and organized by Projekt DEAL.
The authors Leon David Gruenewald and Vitali Koch contributed equally to the work.
Contributor Information
Leon David Gruenewald, Email: leondavid.gruenewald@kgu.de.
Vitali Koch, Email: Vitali.Koch@kgu.de.
Tatjana Gruber-Rouh, Email: tatjana.gruber-rouh@kgu.de.
Axel Thalhammer, Email: axel.thalhammer@kgu.de.
Johannes Frank, Email: johannes.frank@kgu.de.
Ingo Marzi, Email: ingo.marzi@kgu.de.
Christian Booz, Email: boozchristian@gmail.com.
Ibrahim Yel, Email: ibrahim.yel@kgu.de.
Scherwin Mahmoudi, Email: scherwin.mahmoudi@kgu.de.
Simon Bernatz, Email: Simon.Bernatz@kgu.de.
Isabella Laudenberger, Email: isi_laudenberger@hotmail.de.
Neelam Lingwal, Email: Lingwal@med.uni-frankfurt.de.
Thomas J Vogl, Email: T.Vogl@em.uni-frankfurt.de.
Katrin Eichler, Email: katrin.eichler@kgu.de.
REFERENCES
- 1.Kransdorf MJ, Sweet DE. Aneurysmal bone cyst: concept, controversy, clinical presentation, and imaging. AJR Am J Roentgenol 1995; 164: 573–80. doi: 10.2214/ajr.164.3.7863874 [DOI] [PubMed] [Google Scholar]
- 2.Erlemann R, Jundt G. Tumorähnliche knochenläsionen. Radiologe 2016; 56: 507–19. doi: 10.1007/s00117-016-0113-y [DOI] [PubMed] [Google Scholar]
- 3.Lichtenstein L. Aneurysmal bone cyst.A pathological entity commonly mistaken for giant-cell tumor and occasionally for hemangioma and osteogenic sarcoma. Cancer 1950; 3: 279–89. doi: https://doi.org/10.1002/1097-0142(1950)3:2<279::AID-CNCR2820030209>3.0.CO;2-F [Google Scholar]
- 4.Campanacci M, Capanna R, Picci P. Unicameral and aneurysmal bone cysts. Clin Orthop Relat Res 1986; : 25–36. [PubMed] [Google Scholar]
- 5.Bollini G, Jouve JL, Cottalorda J, Petit P, Panuel M, Jacquemier M. Aneurysmal bone cyst in children: analysis of twenty-seven patients. J Pediatr Orthop B 1998; 7: 274–85. doi: 10.1097/01202412-199810000-00005 [DOI] [PubMed] [Google Scholar]
- 6.Oliveira AM, Perez-Atayde AR, Inwards CY, Medeiros F, Derr V, Hsi B-L, et al. Usp6 and CDH11 oncogenes identify the neoplastic cell in primary aneurysmal bone cysts and are absent in so-called secondary aneurysmal bone cysts. Am J Pathol 2004; 165: 1773–80. doi: 10.1016/S0002-9440(10)63432-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sasaki H, Nagano S, Shimada H, Yokouchi M, Setoguchi T, Ishidou Y, et al. Diagnosing and discriminating between primary and secondary aneurysmal bone cysts. Oncol Lett 2017; 13: 2290–96. doi: 10.3892/ol.2017.5682 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hiemcke-Jiwa LS, van Gorp JM, Fisher C, Creytens D, van Diest PJ, Flucke U. USP6-associated neoplasms: a rapidly expanding family of lesions. Int J Surg Pathol 2020; 28: 816–25. doi: 10.1177/1066896920938878 [DOI] [PubMed] [Google Scholar]
- 9.Maiya S, Davies M, Evans N, Grimer J. Surface aneurysmal bone cysts: a pictorial review. Eur Radiol 2002; 12: 99–108. doi: 10.1007/s003300101009 [DOI] [PubMed] [Google Scholar]
- 10.Rapp M, Grauel F, Wessel LM, Illing P, Kaiser MM.. Treatment outcome in 60 children with pathological fractures of the humerus caused by juvenile or aneurysmal bone cysts. Acta OrthopBelg 2016;82(4):723–9. [PubMed] [Google Scholar]
- 11.Grahneis F, Klein A, Baur-Melnyk A, Knösel T, Birkenmaier C, Jansson V, et al. Aneurysmal bone cyst: a review of 65 patients. J Bone Oncol 2019; 18: 100255: 100255. doi: 10.1016/j.jbo.2019.100255 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Urakawa H, Tsukushi S, Hosono K, Sugiura H, Yamada K, Yamada Y, et al. Clinical factors affecting pathological fracture and healing of unicameral bone cysts. BMC Musculoskelet Disord 2014; 15: 159. doi: 10.1186/1471-2474-15-159 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Donaldson S, Wright JG. Recent developments in treatment for simple bone cysts. Curr Opin Pediatr 2011; 23: 73–77. doi: 10.1097/MOP.0b013e3283421111 [DOI] [PubMed] [Google Scholar]
- 14.Sullivan RJ, Meyer JS, Dormans JP, Davidson RS. Diagnosing aneurysmal and unicameral bone cysts with magnetic resonance imaging. Clinical Orthopaedics and Related Research 1999; 366: 186–90. doi: 10.1097/00003086-199909000-00024 [DOI] [PubMed] [Google Scholar]
- 15.Mahnken AH, Nolte-Ernsting CCA, Wildberger JE, Heussen N, Adam G, Wirtz DC, et al. Aneurysmal bone cyst: value of Mr imaging and conventional radiography. Eur Radiol 2003; 13: 1118–24. doi: 10.1007/s00330-002-1668-8 [DOI] [PubMed] [Google Scholar]
- 16.Wootton-Gorges SL. MR imaging of primary bone tumors and tumor-like conditions in children. Magn Reson Imaging Clin N Am 2009; 17: 469–87. doi: 10.1016/j.mric.2009.03.010 [DOI] [PubMed] [Google Scholar]
- 17.Tsai JC, Dalinka MK, Fallon MD, Zlatkin MB, Kressel HY. Fluid-fluid level: a nonspecific finding in tumors of bone and soft tissue. Radiology 1990; 175: 779–82. doi: 10.1148/radiology.175.3.2160676 [DOI] [PubMed] [Google Scholar]
- 18.Wörtler K, Blasius S, Hillmann A, Marx C, Brinkschmidt C, Heindel W. Mr morphology of primary aneurysmal bone cysts: a retrospective analysis of 38 cases. Rofo 2000; 172: 591–96. doi: 10.1055/s-2000-4648 [DOI] [PubMed] [Google Scholar]
- 19.Woertler K. Benign bone tumors and tumor-like lesions: value of cross-sectional imaging. Eur Radiol 2003; 13: 1820–35. doi: 10.1007/s00330-003-1902-z [DOI] [PubMed] [Google Scholar]
- 20.Mankin HJ, Hornicek FJ, Ortiz-Cruz E, Villafuerte J, Gebhardt MC. Aneurysmal bone cyst: a review of 150 patients. J Clin Oncol 2005; 23: 6756–62. doi: 10.1200/JCO.2005.15.255 [DOI] [PubMed] [Google Scholar]
- 21.Vergel De Dios AM, Bond JR, Shives TC, McLeod RA, Unni KK. Aneurysmal bone cyst. A clinicopathologic study of 238 cases. Cancer 1992; 69: 2921–31. doi: https://doi.org/10.1002/1097-0142(19920615)69:12<2921::AID-CNCR2820691210>3.0.CO;2-E [DOI] [PubMed] [Google Scholar]
- 22.Deventer N, Deventer N, Gosheger G, de Vaal M, Vogt B, Budny T. Current strategies for the treatment of solitary and aneurysmal bone cysts: a review of the literature. J Bone Oncol 2021; 30: 100384. doi: 10.1016/j.jbo.2021.100384 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tsukamoto S, Errani C, Facchini F, Papagelopoulos P, Mavrogenis AF. Fluid-fluid levels in musculoskeletal tumor imaging. Curr Med Imaging 2021; 17: 157–65. doi: 10.2174/1573405616666200806173258 [DOI] [PubMed] [Google Scholar]
- 24.Erol B, Onay T, Topkar OM, Tokyay A, Aydemir AN, Okay E. A comparative study for the treatment of simple bone cysts of the humerus: open curettage and bone grafting either without instrumentation or with intramedullary nailing. J Pediatr Orthop B 2017; 26: 5–13. doi: 10.1097/BPB.0000000000000353 [DOI] [PubMed] [Google Scholar]
- 25.Cottalorda J, Kohler R, Sales de Gauzy J, Chotel F, Mazda K, Lefort G, et al. Epidemiology of aneurysmal bone cyst in children: a multicenter study and literature review. J Pediatr Orthop B 2004; 13: 389–94. doi: 10.1097/01202412-200411000-00008 [DOI] [PubMed] [Google Scholar]
- 26.Grzegorzewski A, Pogonowicz E, Sibinski M, Marciniak M, Synder M. Treatment of benign lesions of humerus with resection and non-vascularised, autologous fibular graft. Int Orthop 2010; 34: 1267–72. doi: 10.1007/s00264-009-0911-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Aiba H, Kobayashi M, Waguri-Nagaya Y, Goto H, Mizutani J, Yamada S, et al. Treatment of simple bone cysts using endoscopic curettage: a case series analysis. J Orthop Surg Res 2018; 13(1): 168. doi: 10.1186/s13018-018-0869-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zishan US, Pressney I, Khoo M, Saifuddin A. The differentiation between aneurysmal bone cyst and telangiectatic osteosarcoma: a clinical, radiographic and MRI study. Skeletal Radiol 2020; 49: 1375–86. doi: 10.1007/s00256-020-03432-w [DOI] [PubMed] [Google Scholar]