Six different histopathological diagnoses were identified in this cohort, with chordoma being the most common (48/80 patients, 60.0%), followed by chondrosarcoma (13/80 patients, 16.3%), giant cell tumor of the bone (GCTB) (8/80 patients, 10.0%), plasmacytoma (7/80 patients, 8.7%), osteosarcoma (3/80 patients, 3.7%) and Ewing sarcoma (1/80 patients, 1.3%). We included one dedifferentiated chordoma of the skull base in the chordoma group. One GCTB located in the spine was malignant. One osteosarcoma of the mobile spine was periosteal, and two osteosarcomas located in the mobile spine and sacrum, respectively, were high grade osteosarcomas. Fifty out of 80 patients (62.5%) were male. Median age at diagnosis was 53.5 (range 18–85) with highest median age noted for chordoma (56.5) and lowest for Ewing sarcoma (18). Patient demographics is given in Table 1.
Table 1 Patient demographicsLocation and extensionFrequencies of different diagnoses per location are presented in Table 2. The only two tumor types located in the skull base in our series were chordoma (15/23 patients, 65.2%) and chondrosarcoma (8/23 patients, 34.8%). By far the most common sacral tumor was chordoma (25/36 patients, 69.4%). Other tumors were encountered in the mobile spine and sacrum with rather low frequency.
Table 2 Morphological imaging parameters of 80 patients in six different histopathological groups: chordoma, chondrosarcoma, plasmacytoma, giant cell tumor of the bone (GCTB), osteosarcoma and Ewing sarcomaMost tumors (64/80 patients, 80.0%) originated in the midline. (Table 2) These midline tumors were chordoma (46/64 patients 71.9%), plasmacytoma (6/64 patients 9.4%) and GCTB (6/64 patients 9.4%). At the level of mobile spine and sacrum, there were no major differences between midline and eccentric origin of various tumor types, with the exception of sacral chordoma. All 25 sacral chordomas originated in the midline. In the skull base only 9 tumors were eccentric (petroclival synchondrosis and/or petrous apex), 7 of these were chondrosarcomas (7/9 patients 77.8%). All but one of the midline skull base tumors were chordomas (13/14 patients, 93%).
Soft tissue extension of tumors located in spine or sacrum occurred in 54 patients. It was not seen in three small tumors (one spinal GCTB, one spinal plasmacytoma and one sacral chondrosarcoma). Combined intra- and extraspinal tumor extension was the most common (31/54 patients, 57.4%). (Table 2) Intraspinal or extraspinal extension only was seen in respectively 11 (20.4%) and 12 patients (22.2%). The contribution of chondrosarcomas to the combined intra- and extraspinal extension with only one patient (1/31, 3.2%) was small relative to the overall prevalence of chondrosarcomas. Of note, none of the six patients with plasmacytoma had extraspinal extension only.
Bone destructionGeographical bone destruction (Fig. 3A) was the most frequently seen pattern across all histopathological groups (61/78 patients, 78.2%), while permeated/moth-eaten type of bone destruction (Fig. 3B) occurred in 17 patients (21.8%), predominantly in the mobile spine and sacrum and only once in the skull base. (Table 2) Permeated/moth-eaten type of bone destruction was even the prevailing pattern in chordomas of the mobile spine (6/8 patients, 75.0%), and occurred in a few other tumor types including two of the three sacral chondrosarcomas and the malignant GCTB. This was also the only difference encountered between malignant and benign GCTB within this cohort. Destruction of cortex was common (65/78 patients, 83.3%) in all tumor types regardless of their location.
Fig. 3Examples of patterns of bone destruction: A- Geographical type of bone destruction pattern seen in a skull base chordoma centered dorsally in the clivus (arrows). extending into the apex os petrosum bilaterally. B- Example of permeated/moth-eaten type of bone destruction pattern seen in sacral chordoma (arrows)
CalcificationsRings-and-arcs/popcorn-type calcification was seen on CT in 12 out of 78 patients (15.4%) all diagnosed with chondrosarcoma (Fig. 4A). Amorphous-type calcification (due to pre-existing bone) was seen in 28 patients out of 78 patients (35.9%), all diagnosed with chordoma—12 located in the skull base, 2 in the mobile spine and 14 in the sacrum (Fig. 4B). Dense sclerotic-type calcification was seen in only 3 patients (3.8%), all diagnosed with osteosarcoma (Fig. 4C). In 35 patients (44.9%) no calcification was present. (Table 2).
Fig. 4Examples of different calcification types: A- Axial view CT scan of a skull base chondrosarcoma showing rings-and-arcs/popcorn-like calcification (arrows). B- Axial view CT scan of sacral chordoma showing linear bone remnants (amorphous-like calcification) consistent with pre-existing bone (arrows). C- Axial view CT scan of sacral osteosarcoma presenting with dense sclerotic-like (osteoid) calcification (arrow)
ShapeSmooth margins occurred in the majority of skull base tumors (15/23 patients, 65.2%), especially among chordomas, in accordance with the prevalence of the histological diagnosis. (Table 2) As lobulated margins at the skull base were found in four patients with chordoma (4/8 patients, 50.0%) and in four chondrosarcoma patients, it was thus, relative to the histologic prevalence, more common in chondrosarcomas.
In the mobile spine, smooth margins were most common (13/21 patients, 61.9%), but were not present in the two chondrosarcomas.
Also in the sacrum, smooth margins were most common (21/36 patients, 58.3%), and seen in almost all histological diagnoses. Most patients had, in accordance with the histological prevalence, chordoma. (Fig. 5A) All four patients with GCTB had, as in the mobile spine, smooth margins. Lobulated margins were almost exclusively seen in chordomas (13/15 patients, 86.7%), only one chondrosarcoma and one osteosarcoma patient also had lobulated margins. (Fig. 5B).
Fig. 5Examples of tumor shape. A- Smooth shape seen in a midline located chordoma of the sacrum with intraspinal-extraspinal extension on T2 fat-saturated MR image. B- Lobulated shape seen in an eccentrically located chondrosarcoma of the mobile spine with intraspinal-extraspinal extension on T2 fat-saturated MR image
Cystic componentsCystic components were seen in only five out of 80 patients (6.3%). (Table 2) A single cystic component was seen in one patient with chordoma of the mobile spine. Multicystic changes were depicted in two patients with sacral GCTB and two with osteosarcoma. The cystic changes in the two sacral GCTB also had fluid–fluid levels consistent with ABC-like changes.
Signal Intensity (SI)Predominantly hyperintense areas were visible on T2-weighted MR sequences in the majority of patients (74/80 patients, 92.5%), independent of histology and location. (Table 2) These areas appeared heterogeneous in 49 patients (Fig. 6A); distributed over chordomas (39/49 patients, 79.6%), GCTB (4/49 patients, 8.2%), chondrosarcoma (3/49 patients, 6.1%), and one each of osteosarcoma, plasmacytoma and Ewing sarcoma. These hyperintense areas appeared homogenous in the remaining 25 patients who had plasmacytoma (5/25 patients, 20.0%), chondrosarcoma (9/25 patients, 36.0%) (Fig. 6B), chordoma (9/25 patients, 36.0%) or GCTB (2/25 patients, 8.0%). Hypointense areas on T2-weighted sequences were seen in five tumors only; two of these were GCTB (Fig. 6C), and one each was chondrosarcoma, osteosarcoma and plasmacytoma. Isointensity was seen in only one patient with osteosarcoma.
Fig. 6Examples of signal intensity. A- Heterogenous hyperintensity seen on T2-weighted MRI sequence in a midline located chordoma of the sacrum with extraspinal extension. B- Homogeneous hyperintensity seen on T2-weighted MRI sequence in chondrosarcoma of the skull base. C- Hypointensity seen on T2-weighted MRI sequence in a midline located sacral giant cell tumor of the bone with intraspinal extension
On T1-weighted sequences hypointense areas were seen in 41 patients (51.3%) of which 24 had chordoma, 5 chondrosarcoma, 8 GCTB, 2 osteosarcoma, 1 plasmacytoma and 1 Ewing sarcoma. Hyperintense areas were seen in 32 patients (40.0%), out of which 19 had chordoma, 7 chondrosarcoma and 6 plasmacytoma. Isointense areas were seen in only 7 patients (8.7%) out of which 5 had chordoma, one chondrosarcoma and one osteosarcoma.
Contrast Enhancement (CE)Skull base tumors showed predominantly high CE (61–100%) independent of histology. In the mobile spine this was also the most common enhancement among all histologies, with the exception of the two chondrosarcomas who both displayed low-intermediate CE (1–61%). (Table 2) In the sacrum, low-intermediate CE was by far the most frequent in chordomas (15/18 patients, 83.3%). Most other tumors, with the exception of chondrosarcomas and osteosarcomas showed high CE (61–100%).
Reticular CE pattern (Fig. 7A) was most commonly seen in chordoma patients, septo-nodular CE pattern (Fig. 7B) in patients with chondrosarcoma of the mobile spine and sacrum, and homogeneous CE pattern in patients with skull base chondrosarcoma (Fig. 7C) as well as in patients with plasmacytoma of the mobile spine and sacrum (Fig. 7D). Inhomogeneous CE (Fig. 7E) was seen in patients with chordoma, GCTB, osteosarcoma and Ewing sarcoma.
Fig. 7Examples of different types of contrast enhancement: A – reticular enhancement pattern in a midline located clival chordoma. B – septo-nodular enhancement pattern in an eccentrically located chondrosarcoma of the mobile spine with intraspinal-extraspinal extension. C – homogeneous enhancement type in chondrosarcoma located at the midline of the petroclival junction. D – homogeneous enhancement type in an eccentrically located sacral plasmacytoma with intraspinal-extraspinal extension. E – inhomogeneous enhancement type in a midline located sacral chordoma with intraspinal-extraspinal extension
MR perfusionPerfusion analysis was performed in 73 patients. (Table 3) Maximal enhancement values of chordoma were significantly lower compared to those of chondrosarcoma (p = 0.046), GCTB (p = 0.044) and plasmacytoma (p = 0.005). Maximal relative enhancement (p = 0.009) and wash in rate values (p = 0.003) were significantly lower in chordoma compared to plasmacytoma. Maximal relative enhancement was also significantly lower (p = 0.013) in chordoma than GCTB. Time to peak values were significantly longer in both chordoma and chondrosarcoma compared to GCTB (p = 0.003; p = 0.004) and plasmacytoma (p = 0.004; p = 0.002). Wash in rate values of chondrosarcoma were significantly lower compared to plasmacytoma (p = 0.025).
Table 3 Mean values of perfusion parameters per histopathological diagnosis. Average area of the ROI – 93,50 mm2Of the TIC’s, Type I was only seen in six patients with chordoma, and in no other tumor. (Fig. 8A) Type II curve was only seen in chordoma (9/14 patients, 64.3%), and chondrosarcoma (5/14 patients, 35.7%) (Fig. 8B). Type III curve was the most common one in chordoma (17/26 patients, 65.4%) (Fig. 8C), and GCTB (6/26 patients, 23.1%). It was also seen in two osteosarcomas, and one chondrosarcoma. Type IV curve was the only curve type seen in plasmacytoma (5/10 patients, 50.0%) (Fig. 8D); it was also encountered with lower frequencies in GCTB (2/10 patients, 20.0%), chordoma (2/10 patients, 20.0%), and chondrosarcoma (1/10 patients, 10.0%). Type V curve was seen in chondrosarcoma (5/17 patients, 29.4%) (Fig. 8E), chordoma (11/17 patients, 64.7%) and Ewing sarcoma (1/17 patients 5.9%).
Fig. 8Time intensity curve (TIC) of the lesion and artery measured within the circular ROIs. A—Curve type I seen only in 8.2% of the patients, all diagnosed with chordoma – Example of a clival chordoma TIC. B—Curve type II seen in 19.2% of the patients, mostly chondrosarcoma and chordoma – Example of clival chondrosarcoma TIC. C—Curve type III seen in 35.6% of the patients, mostly chordoma, GCTB, osteosarcoma and chondrosarcoma—Example of clival chordoma TIC. D—Curve type IV seen in 13.7% of the patients, mostly plasmacytoma, including GCTB, chordoma and chondrosarcoma—Example of sacral plasmacytoma TIC. E—Curve type V seen in 23.3% of the patients, mostly chondrosarcoma, chordoma and Ewing sarcoma – Example of clival chondrosarcoma
MR permeabilityPermeability parameters were analyzed in 57 patients. (Table 4) Although Vp of chondrosarcoma was five times larger than that of chordoma, the parameters of chordoma and chondrosarcoma were not significantly different. However, Ktrans, Kep and Vp were significantly lower in chordoma (p = 0.011; p = 0.004; p = 0.003) and chondrosarcoma (p = 0.014; p = 0.004; p = 0.007) compared to those of plasmacytoma. Values of Ktrans and Kep were significantly lower in chordoma ( p = 0.010; p = 0.011) and chondrosarcoma (p = 0.010; p = 0.006) compared to GCTB.
Table 4 Mean values of permeability parameters per histopathological diagnosis. Average area of the ROI—96,64 mm2Diffusion-weighted MRIOverall chondrosarcoma had significantly higher ADCmin (p = 0.008), ADCmax (p = 0.017) and ADCmean (p = 0.008) values compared to chordoma. (Table 5) Location wise this difference was most significant among the skull base tumors with p-values of < 0.001, 0.002and < 0.001 for ADCmin, ADCmax and ADCmean, respectively. Single available cases of osteosarcoma, Ewing sarcoma, GCTB and plasmacytoma had lower ADCmean values compared to chondrosarcoma—1.905·10−3mm2/s, 0.985·10−3mm2/s, 1.151·10−3mm2/s and 0.992·10−3mm2/s, respectively.
Table 5 Mean ADC values of CH and CS per location in 10–3 mm2/s. Average area of the ROI – 68.29 mm2Subjective final diagnosisOverall, differential radiological diagnosis considered by the two reviewing radiologists, based on the morphological imaging parameters, coincided with the histopathological diagnosis in 72 patients (90%). In three patients (3.75%), two with chordoma and one with osteosarcoma, erroneous diagnoses of hemangioma, chondrosarcoma and GCTB were made. In 5 patients (6.25%) an unequivocal diagnosis could not be made. In two chordoma and one chondrosarcoma patients unequivocal diagnosis between chordoma and chondrosarcoma could not be made and, in two GCTB patients GCTB could not be differentiated from plasmacytoma.
Upon reviewing the functional imaging parameters of misdiagnosed patients, only 3 chordomas falsely diagnosed as chondrosarcoma could be considered as chordoma based on perfusion wash in rate parameter (21.87 s−1, 30.05 s−1, 50.84 s−1) and permeability Vp parameter (2.71·10–3, 7.09·10–3, 2.62·10–3) values which were approximate to the median values for chordoma presented in Tables 3 and 4.
Comments (0)