MRI in primary bone tumors: therapeutic implications

European Journal of Radiology. 12 (1991) 201-207

201

Elsevier

EURRAD

00143

MRI in primary bone tumors: therapeutic implications R. Golfieri

’ 3*,

H. Baddeley

‘,

J. S. Pringle 2, A. W.L. Leung ‘, A. Greco’

and R. Souhami2

’ The Paul Strickland Scanner Centre. Mount Vernon Hospital, Northwood and ’ The London Bone Tumour Service, The Institute of Orthopaedic’s, London,

(Received

Key words: Bone neoplasm,

13 August 1990; accepted

MRI; Bone neoplasm,

U.K.

after revision 16 November

staging; Magnetic resonance

1990)

imaging, comparative

study

Abstract

The accuracy of preoperative MRI in detecting tumor extent has been evaluated in 35 patients with primary bone neoplasms; intra-osseous extent was measured on MR images and compared with macroslides of surgical specimens in 26 cases. An almost completely accurate prediction of tumor size was obtained with the combined employment of Spin-Echo (SE) and Short Inversion Time Inversion Recovery (STIR) sequences in the various tumors, with the exception of two Ewing’s sarcomas. Changes in Signal Intensity (SI) and tumor morphology were identified in those cases which had undergone presurgical chemotherapy: the reduction in SI and in tumor size or the appearance of a more homogeneous signal was correlated with a positive response to cytotoxic therapy. MR imaging fully satisfies surgeon’s preoperative requirements in the assessment of therapy-responding neoplasms as well as in local tumor staging in all types of neoplasms, with the exception of Ewing’s sarcoma.

Introduction MRI is the most reliable imaging method in the presurgical staging of bone neoplasms because of its high accuracy, compared to conventional radiology, CT and nuclear medicine, in detecting intramedullary tumor extent, in depiction of skip lesions and extraperiosteal soft-tissue spread [ 1,2,20,21,24-26,28,30]. An accurate assessment of the longitudinal extent of the tumor and identification of discontinuous tumor is of critical importance for planning the limits of resection in conservative surgery [ 9,10,18]. Some studies have been reported in the literature comparing the results of intramedullary extent measurements as appearing on MR and in surgical specimens of osteosarcomas [ 13,231 or in limited series [4,28]. To our knowledge there have been no reports referring to a wide range of tumors. The decision to plan limb-saving resection as opposed to amputation is based in part on the response to presurgical chemotherapy employed in order to * Dr Rita Gollieri was on a leave of absence from the III Department of Radiology,

S. Orsola University

Hospital,

Bologna, Italy. D’Urgenza, 9, 40138 Bologna, Italy.

Addressfir

reprints: Dr. Rita Golfieri, Radiodiagnostica

Policlinico

S. Orsola, Via Massarenti

0720-048X/91/$03.50

0 1991 Elsevier Science Publishers

B.V.

destroy distant micrometastases. The degree of tumor necrosis, expressing cell sensitivity to treatment and more likely involving distant metastases, was found to be of great prognostic value with respect to patient survival, particularly in Ewing’s and osteogenic sarcoma [ 9,10,18]. The prediction of good response to treatment is therefore crucial in surgical planning: the MR appearances of chemotherapy changes have been reported, with conflicting results [3,17,28,29]. We evaluated the accuracy of MRI (0.5 T) in measuring the intraosseous extent in various bone tumors by comparing MR images with macroslides of postresection specimens in 26 patients. Fifteen cases which underwent presurgical chemotherapy were also studied in order to detect distinctive SI patterns or morphological changes indicating a positive response to treatment. Patients and Methods 35 MRI studies for primary bone tumors performed in 26 patients were reviewed. The histopathologic diagnoses and localizations are presented in Table 1. All patients were surgically treated within 10 days after MRI examination and a comparison between MRI

202 TABLE 1 26 MRI studies compared

with surgical macroslides

Tumor type

Localization

No chemotherapy: 1 benign chondroblastoma 4 chondrosarcomas 2 osteochondromas 2 parosteal osteosarcomas 2 giant-cell tumors

humerus( 1) tibia(2) femur(2) tibia(l) femur(l) femur (2) femur (2)

MRI pre + post chemotherapy: 5 osteosarcomas 4 Ewing’s sarcomas

femur (1) tibia(2) humerus (1) fibula( 1) femur(2) ulna(2)

MRI post-chemotherapy 4 osteosarcomas 2 Ewing’s sarcomas

femur (4) femur (2)

results and the macroslides of the surgical specimen was performed. Fifteen patients of this group had undergone presurgical chemotherapy of which nine had MRI performed both before and after chemotherapy (Table 1) and a comparison in tumor morphology and extent before and after treatment was carried out. All MRI examinations were performed with a Picker International 0.5 Tesla Superconducting Body Scanner operating at 21.3 MHz. In all the patients, MR Spin Echo (SE) scans were performed using short Repetition Time (TR)/Echo Time (TE) of 7-800/26/2-4 (TR/TE/N of repetitions) (i.e., ‘Tl-weighted’) and long TR/TE of 2000/80/2-4 (i.e., T2-weighted). Short Inversion Time Inversion Recovery (STIR) sequence was also employed in all examinations, using a TR of 1800-2400 ms and an Inversion Time (TI) of 120-130 ms: this value was selected for the purpose of suppressing the signal of fat [ 6,7]. At least two imaging modalities were used in each MR examination and the STIR sequence was always performed in the same plane of one SE sequence, in order to obtain a direct comparison between them. All the macroslides of the surgical specimens were prepared from the anatomical piece, sliced in the longitudinal plane better representing the tumor extent, including the entire tumor and a large amount of normal bone surrounding the lesion, or the adjacent articular surface. The measurement of the intraosseous neoplastic extent was carried out as follows: the complete border of the lesion was marked with a line-tipped pen and the major tumor extent parallel to the longitudinal axis was then measured. The measurement of each tumor was linear and not volumetric. On MR images, intra-osseous tumor extent measurements were obtained using as a reference point

a Centimeter scale recorded on each film. Only MR images performed on the longitudinal plane (i.e., coronal or sagittal) were used for measurements. In all cases the STIR sequence was used for measurements. Among SE pulse sequences, short TE/TR (Tlweighted) was considered for intra-osseous measurements, as the method of choice in detecting medullary invasion [ 8,301; conversely long TE/TR (T2-weighted) was employed for the assessment of soft-tissue involvement [5]. Direct measurement of the entire soft-tissue component was unobtainable on the surgical specimens because it was fragmented at surgical cut: the presence of extraperiosteal soft tissue spread was indirectly deduced from the presence of periosteal breach. The Signal Intensity (SI) was evaluated on SE sequences as follows: high (i.e., same intensity as surrounding fat), intermediate (i.e., SI of muscle) low (’i.e., SI of intact cortical bone). In assessing postchemotherapy changes, an increase or decrease in SI was evaluated comparatively to the SI of the reference tissue. Results Comparison between MR images and surgical macroslides

The mean intra-osseous tumor extent measured on macroslides, was 7.7 2 3.5 cm. The mean differences between the tumor extent measured on macroslide and that on MR scans was 0.57 & 07 cm. MR measurements corresponded precisely to macroslides in 5 cases (19%), overestimated tumor extent in 7 (35%), whereas a slight underestimation of longitudinal extent was present in 12 cases (46%). No correlation has been found between measurements discrepancy and tumor types, only exception being Ewing’s sarcomas studied after chemotherapy, in which a marked underestimation of tumor extent was always detected by MR studies (mean difference MR/macroslides 2.5 cm). This finding was related to the peculiar infiltrative growth pattern of this neoplasm, and to the presence of an admixture of bony sclerotic reaction and proliferating cells at the tumor periphery, causing an indistinct low SI on STIR and T2-weighted studies. In one of these, the STIR sequence detected high SI red bone marrow (Fig. 1) in both femoral diaphyses, thus rendering tumor detection more difficult. When the two cases of Ewing’s sarcoma are excluded from calculations, the discrepancy between MR and lowered to measurements was macroslides 0.3 f 0.3 cm. In two cases ofgiant cell tumors (GCT), a subtle high SI ‘veiled’ appearance was detected by the STIR

203

Fig. I. Ewing’s sarcoma of the right femoral diaphysis; 9-year-old male. MRI performed after three successful courses of chemotherapy. (a, b and c) Coronal 120/2000 STIR: poor definition of longitudinal tumor limits. No clear demarcation from normal SI bone marrow. Slight diffuse and symmetrical increase in SI in both femora (arrows) due to high content of red bone marrow. (d) Surgical specimen: minimal trabecular destruction and diffuse sclerotic bone reaction.

sequence in perilesional bone marrow: this finding, related to hyperemic medullary response to the hypervascular nature of GCT [ 151, was always differentiated from tumor. On measuring intra-osseous extent, STIR provided a more accurate depiction of tumor limits than the Tl-weighted sequence in 19% of cases (Fig. 2). In the detection of subperiosteal spread (Fig. 2) or periosteal breakthrough, a perfect correlation with macroslide findings was obtained with the combined use of T2-weighted and STIR sequences. On comparison of the two sequences, STIR showed a higher accuracy than T2-weighted in defining the real extraperiosteal extent in 5 cases [ 19% 1. In 16 cases of malignant tumors (61.5x), STIR showed a high SI, ‘veiled’ appearance of uninvolved muscles : the distinction between tumor infiltration and

reactive or ischemic compressive possible in all cases [ 141. Post-chemotherapy

edema was, however,

changes

In 4 osteosarcomas and 6 Ewing’s sarcomas which showed a positive response to treatment, a reduction in both tumor SI and size was noted on T2-weighted and STIR sequences (Fig. 3). Two osteogenic and two Ewing’s sarcomas of this responding group developed a pseudocapsule of new sclerotic bone surrounding the intramedullary extent. This low SI peripheral halo in the Ewing’s sarcoma caused an underestimation of neoplastic extent due to the difficulty in identifying tumornormal marrow interfaces (Fig. 1). In two osteosarcomas previously showing a higher SI in the more cellular extra-osseous portion, a low SI necrotic response was obtained mainly in the peripheral portion,

204

and the MR signal became as isointense as the endosteal portion. In 5 cases pretreatment MR scans showed soft-tissue edema which disappeared after successful chemotherapy. The decrease in peritumoral edema was considered a good response marker in these cases. In five osteosarcomas negative therapeutic response was expressed by an increase in tumor SI and size, with the appearance of more marked inhomogeneities, multiple high SI areas in the intra-osseous portion, due to extensive hemorrhagic necrosis. An increase of muscle edema also appeared on T2weighted and STIR images. Discussion In planning limb-salvage resections the data needed by the orthopedic surgeons are (a) an accurate local

tumor staging and (b) the knowledge of a necrotic tumor response to presurgical chemotherapy [lo]. In evaluating tumor response to chemotherapy, in our series a good prognosis was usually connected with the decrease of SI on T2-weighted and STIR sequences, usually associated with a decrease in tumor size [3,23,28]. Lower SI values were related to necrotic changes and increased cellular ‘maturation’ with production of fully calcified bone within the tumor. The sclerotic bone reaction is mostly peripheral, with the appearance of a pseudocapsule [ 171: in osteosarcoma it tends to create a sharper demarcation of tumor limits (Fig. 3), whereas in Ewing’s sarcoma it gradually shades into normal bone, sometimes obscuring tumor margins (Fig. 1). We found the STIR sequence highly sensitive in detecting soft-tissue edema [ 141: following successful

205

Fig. 3. Right tibia high grade conventional osteosarcoma: good response to chemotherapy. (a) Pre-chemotherapy coronal 80/2000 SE and (b) 120/2200 STIR Sequences. (c) Post-chemotherapy long SE and (d) STIR sequences: reduction in SI and size of the extra-osseous component, mainly in its peripheral, cellular portion. (e) Surgical macroslide, coronal plane: precise correlation of tumor extent.

206

chemotherapy we observed the complete disappearance of the high signal from muscle planes and this could be considered a further marker of good response to treatment. In tumors showing increased SI on the post-treatment scan, histopathology demonstrated a progression in tumor cellularity, hemorrhagic necrosis or reactive edematous changes. These findings were always connected to a slight progression in tumor size and therefore to poor response to cytotoxic therapy. Our data are therefore in contrast with previous studies [ 3,291, indicating that an increase in SI on T2-weighted sequences represents a good response to chemotherapy. In our study, an accurate prediction of intra-osseous extent was always obtained with the combined use of Tl-weighted and STIR sequences, with the only exception of two Ewing’s sarcomas studied in a post-chemotherapy phase, in which both sequences underestimated the intramedullary extent. Therefore, staging of Ewing’s sarcoma requires careful attention because it seems to be the only tumor prone to underestimation on MRI studies, due to its subtle infiltrative growth pattern [22] and its sclerotic bone response to cytotoxic therapy [4]. In all cases intramedullary neoplastic extent was better depicted by the STIR sequence which more precisely delineates tumor-normal marrow interfaces, by nulling the signal of fatty marrow [ 161. The STIR sequence also produces a high signal response from tissue with long Tl and T2 relaxation times. This peculiarity increases its sensitivity in tumor detection [ 6-8,12,14] but decreases its specificity: tumor, inflammation and peritumorous edema all show a high SI on this pulse sequence. In accordance with a previous report [ 131, no histologic evidence of peritumoral bone marrow edema was found in our series; this finding can explain the high accuracy of STIR in the detection of intra-osseous extent. The STIR sequence depicted a slight diffuse increase in SI in bone marrow in two cases of GCT and in one Ewing’s sarcoma (Fig. 1). However, these findings, which could potentially overstage tumors, were easily recognized as typical marrow hyperemic response [ 151 in the former and as the common appearance of red bone marrow in young patients in the latter [ 11,191. Extra-osseous tumor extent was precisely defined in all cases by comparing the results of TZweighted and STIR scans. The STIR sequences showed higher lesion-to-background contrast than TZ-weighted sequence in all cases, but also provided an apparent overstaging due to high SI peritumoral edema. This finding was always differentiated from tumor, based on morphology of the abnormal areas [ 151. On comparing T2-weighted and STIR sequences, the former has been

the most suitable in detecting gross morphologic details, when an attempt in tumor characterization is needed, whereas for an accurate staging the STIR sequence appears the most sensitive in depicting the whole neoplastic extent both in its intramedullary and extraosseous components, as confirmed by the comparison with surgical macroslides in our series. References 1 Aisen AM, Martel W, Braunstein EM, McMillin KI, Phillips WA, Kling TF. MRI and CT evaluation ofprimary bone and soft tissue tumors. AJR 1986; 146: 749-756. 2 Berquist TH. Bone and soft tissue tumors. In Berquist, Ehman, Richardson, eds. Magnetic Resonance of the Musculoskeletal System. New York, Raven Press, 1987: 85-108. 3 Bloem JL, Bluemm RG, Taminiau AHM, Van Oosterom AT, Stolk J, Doornbos J. Magnetic resonance imaging of primary malignant bone tumors. Radiographics 1987; 7: 425-445. 4 Bloem JL, Taminiau AHM, Eulderink F, Hermans J, Pauwels EKJ. Radiologic staging of primary bone sarcoma. MR imaging, scintigraphy, angiography and CT correlated with pathologic examination. Radiology 1988; 169: 805-810. 5 Boyko OB, Cory DA, Cohen MD, Provisor A, Mirkin D, DeRosa GP. MR imaging of osteogenic and Ewing’s sarcoma. AJR 1987; 148: 317-322. 6 Bydder GM, Young IR. MR imaging: clinical use of Inversion Recovery sequence. J Comput Assist Tomogr 1985; 9: 659-675. 7 Byddh GM, Steiner RE, Blumgart LH, Khenia S, Young IR. MR of the liver using short TI Inversion Recovery Sequences. J Comput Assist Tomogr 1985; 9: 1084-1089. 8 Burk DL Jr, Dalinka MK, Schiebler ML, Cohen EK, Kressel HY. Strategies for musculoskeletal magnetic resonance imaging. Rad Clin North Am 1988; 26: 653-672. 9 Campanacci M. Surgical treatment: margins and reconstructions. Presented at the Consensus meeting on Osteosarcoma: state of art in Europe. Bologna, Italy, April 9-10, 1987. 10 Chawla S, Benjamin R, Jaffe N et al.. Preoperative intraarterial cisplatin and limb salvage surgery for patients with high-grade osteosarcoma of the extremities. In Salmon, ed., Adjuvant therapy of cancer. Vol 5, New York: Grune & Stratton, 1987: 701-710. 11 Daffner RH, Lupetin AR, Dash N, Deeb ZL, Sefczek RJ, Schapiro RL. MRI in the detection of malignant infiltration of bone marrow. AJR 1986; 146: 353-358. 12 Dwyer AJ, Frank JA, Sank VJ, Reinig JW, Hickey AM, Doppman JL. Short TI inversion-recovery pulse sequence: analysis and initial experience in cancer imaging. Radiology 1988; 168: 827-836. 13 Gillespy T III, Manfrini M, Ruggieri P, Spanier SS, Pettersson H, Springfield DS. Staging of intraosseous extent of osteosarcoma: correlation of preoperative CT and MR imaging with pathologic macroslides. Radiology 1988; 167: 765-767. 14 Goltieri R, Baddeley H, Pringle JS, Souhami R. The role of the STIR sequence in MRI examination of bone tumours. Br J Radio1 1990; 63: 251-256. 15 Gollieri R, Baddeley H, Pringle JS, Leung WL, Greco A, Souhami R. Primary bone tumors: MRI morphologic appearance correlated with pathologic examinations (submitted for publication). 16 Haggar AM, Froelich JW. MR Imaging strategies in primary and

207

metastatic malignancy. Rad Clin North Am 1988; 26: 689-696. 17 Kanal E, Burk DL JR, Brunberg JA, Johnson ND, Wood BP, Flom L. Pediatric musculoskeletal magnetic resonance imaging. Rad Clin North Am 1988; 26: 211-239. 18 Kemp H. Limb conservation surgery for osteosarcoma and other primary bone tumors. Bailliere’s Clin Oncol 1987; 1: 116-136. 19 Kricun ME. Red-yellow marrow conversion: its effect on the location of some solitary bone lesions. Skeletal Radio1 1985; 14: 10-19. 20 Moser RP, Madewell JE. An approach to primary bone tumors. Rad Clin North Am 1987; 25: 1046-1093. 21 Pettersson H, Gillespy T III, Hamlin DJ et al. Primary musculoskeletal tumors: examination with MR imaging compared with conventional modalities. Radiology 1987; 164: 237-241. 22 Pringle JAS. Pathology of bone tumors. Bailliere’s Clin Oncol 1987; 1: 21-63. 23 Redmond OM, Stack JP, Dervan PA, Hurson BJ, Carney DN, Ennis JT. Osteosarcoma: use of MR imaging and MR spectroscopy in clinical decision making. Radiology 1989; 172: 811-815. 24 Reiser M, Rupp N, Biehl TH et al. MR in the diagnosis of bone turnouts. Eur J Radio1 1985; 5: l-7.

25 Richardson ML, Kilcoyne RF, Gillespy T III, Helms CA, Genant HK. Magnetic resonance imaging of musculoskeletal neoplasms. Rad Clin North Am 1986; 24: 259-267. 26 Sartoris DJ, Resnick D. MR Imaging of the musculoskeletal system: current and future status. AJR 1987; 149: 457-467. 27 Shuman WP, Baron RL, Peters MJ, Tazioli PK. Comparison of STIR and spin-echo MR imaging at 1.5 T in 90 lesions of the chest, liver and pelvis. AJR 1989; 152: 853-859. 28 Sundaram M, McGuire MH, Herbold DR, Wolverson MK, Heiberg E. Magnetic resonance imaging in planning limbsalvage surgery for primary malignant tumors of bone. J Bone Joint Surg 1986; 68: 809-819. 29 Wetzel LH, Levine E, Murphey MD. A comparison of MR imaging and CT in the evaluation of musculoskeletal masses. Radiographics 1987; 7: 851-874. 30 Zimmer WD, Berquist TH, McLeod RA et al. Bone tumors: magnetic resonance imaging versus computed tomography. Radiology 1985; 155: 709-718.