Blood Perfusion of Vertebral Lesions Evaluated With Gadolinium-Enhanced Dynamic MRI: In Comparison With Compression Fracture and Metastasis

Transcription

1 Original Research JOURNAL OF MAGNETIC RESONANCE IMAGING 1: (2002) DOI /jmri Blood Perfusion of Vertebral Lesions Evaluated With Gadolinium-Enhanced Dynamic MRI: In Comparison With Compression Fracture and Metastasis Wei-Tsung Chen, 1 Tiffany Ting-Fang Shih, 2 * Ran-Chou Chen, 1 Hsin-Yen Lo, 3 Chen-Te Chou, 1 Jiunn-Ming Lee, 1 and Hsing-Yang Tu 1 Purpose: To investigate blood perfusion of vertebral lesions using dynamic Gd-DTPA-enhanced MRI. Materials and Methods: Dynamic MR studies were performed for cases of acute compression fracture, chronic compression fracture, metastatic vertebral lesions with or without compression fracture. A total of 2 patients (71 vertebral segments) were included. Five types of time-intensity curves (TICs) were defined as: nearly no enhancement (type A), slow enhancement (type B), a rapid contrast wash-in followed by an equilibrium phase (type C), a rapid contrast wash-in followed by early wash-out (type D), and a rapid contrast wash-in with a second slower-rising slope (type E). Results: Metastatic vertebral lesions with or without fracture had a higher peak enhancement percentage and steeper enhancement slope than those of chronic compression fracture, but had no difference as compared to those of acute compression fracture. The type D curve had high positive predictive value for metastatic group (100%), and the type E curve had high positive predictive value for benign compression fracture (8.7%). Conclusion: Type D and E curves are valuable in the differentiation of benign and malignant vertebral lesions. Key Words: Gd-DTPA-enhanced dynamic MR; time-intensity curve; compression fracture; metastasis; pathologic fracture J. Magn. Reson. Imaging 2002;1: Wiley-Liss, Inc. GADOLINIUM diethylenetriamine-pentaacetic acid (Gd- DTPA) shows pharmacokinetics similar to that of iodinized contrast media, and produces an increase of signal 1 Department of Radiology, Taipei Municipal Jen-Ai Hospital, Taipei, Taiwan. 2 Department of Medical Imaging, National Taiwan University Hospital, Taipei, Taiwan. 3 Department of Radiology, Taipei Municipal Chung-Shau Hospital, Taipei, Taiwan. *Address reprint requests to: T.T.-F.S., Department of Radiology, National Taiwan University, Medical College and Hospital, No.7, Chung- Shan S Rd. Taipei, Taiwan. ttfshih@ha.mc.ntu.edu.tw Received November 1, 2000; Accepted November, intensity (SI) in the T1-weighted spin-echo magnetic resonance (MR) image by reducing T1 relaxation time (1). It has been demonstrated that zones showing a marked increase in SI are correlated with highly vascularized regions, whereas zones with low or no increase in SI are often correlated with necrotic area in malignant tumors (2,3). In recent years, dynamic MRI has been used for evaluating musculoskeletal neoplasms () and monitoring response to chemotherapy (). In 199, Verstraete et al (6) concluded that first-pass data from dynamic contrast-enhanced MRI depicted tissue vascularization and perfusion of musculoskeletal neoplasms, although an overlap in the slope values was demonstrated in high vascular benign and malignant lesions. Van Der Woude et al (7) also utilized dynamic MRI to detect residual viable tumor of postchemotherapy osteosarcomas and Ewing s sarcoma before surgery. In 1997, Bollow et al (8) used dynamic MRI to compare normal and malignant bone marrow infiltrations. Thus, dynamic MRI proved to be an effective and noninvasive method for evaluating the in vivo blood perfusion of bone, marrow, and tumors. MRI is highly useful for detecting diseases of the bone marrow, and has become the imaging modality of choice for marrow metastatic disease (9 11). However, differential diagnosis of vertebral compression fracture without obvious clinical or radiologic evidence of malignancy in an elderly patient presents a difficult problem (12 1). To our knowledge, there have been a limited number of articles evaluating bone marrow blood perfusion in normal or diseased vertebrae (16,17). Bone marrow perfusion evaluated with gadolinium-enhanced dynamic MRI was reported by Cova et al (16) in 1991 and showed a strong correlation between MR data and microsphere blood-flow measurements. Therefore, our study investigated the peak contrast enhancement percentage, enhancement slope, and time-intensity curve (TIC) patterns in the first pass of contrast into diseased vertebral bodies to analyze the perfusion characters of compression fracture and metastatic vertebrae with and without pathologic fractures Wiley-Liss, Inc. 308

2 Blood Perfusion of Vertebral Lesions 309 MATERIALS AND METHODS Patients Forty-two patients (16 males and 2 females, 39 8 years old [mean, 66. years]) were included in this study. They underwent contrast-enhanced dynamic MRI as part of the MRI assessment of spinal lesions in this prospective study. The patients were divided into four groups and their MR images were assessed as follows. Group I consisted of 11 patients (two males and nine females, 0 82 years old [mean, 70.9 years]) with 12 acute compression fractures of vertebral bodies. Acute compression was defined as a 1-month symptom duration (as determined by clinical history) before MR examination (13). Group II consisted of 17 patients (three males and 1 females, 60 8 years old [mean, 7.1 years]) with 21 chronic compression fractures of vertebral bodies, with disease duration of 1 month. The patients in groups I and II all had at least 6 months of follow-up without evidence of neoplasm or malignancy of the vertebrae. Group III consisted of 17 patients who had known metastatic disease to the vertebral bodies (11 males and six females, 39 7 years old [mean, 8 years]). Thirty-two cases of non-collapsed metastatic vertebrae were included in this study. The primary neoplasms of these patients included hepatocellular carcinoma (N 3), esophageal cancer (N 1), breast cancer (N 3), rectal carcinoma (N 1), lung cancer (N ), renal cell carcinoma (N 1), multiple myeloma (N 2), and uterine carcinoma (N 1). Three patients had pathological confirmation from biopsy of involved vertebrae. The two multiple myeloma cases were confirmed by bone marrow aspiration. Four patients had multiorgan metastasis and received pathologic confirmation from other metastatic sites (liver, ovary, supraclavicular lymph node, and subcutaneous metastatic nodule). In the remaining eight cases, the diagnosis of metastatic disease was based on the clinical skeletal presentations, underlying malignancy, and characteristic findings from radionuclide bone scintigraphy. MRI in pre- and postcontrast studies presented abnormal intramedullary lesions within the involved vertebra and showed progressive change in follow-up studies. Six patients (five males and one females, 71 years old [mean, 6 years]) in group III had six segments of pathologic vertebral fracture, which were classified as group IV. The primary neoplasms were hepatocellular carcinoma (N 1), breast cancer (N 1), rectal carcinoma (N 1), lung cancer (N 2), and multiple myeloma (N 1). MRI MRI was performed on a 1.-Tesla superconducting system (Powertrak 6000; Philips, The Netherlands). A synergy spine coil was used for the thoracolumbar region. Spin-echo T1-weighted images (repetition time (TR) 0 msec, echo time (TE) 12 msec), turbo spin-echo T2 images (TR 2800 msec, TE 110 msec) with or without a fat suppression technique were performed for the sagittal and axial views before contrast administration with a 30- and 20-cm field of view, respectively. The slice thickness was mm. The contrastenhanced dynamic MR pulse sequence used ultrafast T1-weighted gradient-echo sequences (turbo-field echo; Philips) with TR.6 msec, TE 1. msec, pre-pulse inversion time 00 msec, flip angle 1, scan percentage 70%, and acquisition matrix The slice thickness was 10 mm, placed in the midsagittal plane of the vertebral body. A total of 120 dynamic images were obtained in 83 seconds. A bolus of dimeglumin gadopentetate (Magnevist; Schering, Berlin, Germany) (0.1 mmol/kg/body weight) was rapidly administered manually via a previously placed 21-gauge intravenous catheter in the right antecubital vein, immediately followed by a 20-mL saline flush at the same injection rate. The dynamic scan started as soon as the injection of the contrast medium commenced. Data Analysis SI values were measured in operator-defined regions of interest (ROIs) over the abnormal signal area in T1- weighted images using a cursor and graphic display device (Fig. 1a). The SI values derived from the ROIs were plotted against time (Fig. 2) as TIC using the Gyroview software system (Philips). The baseline value (SI baseline ) of the SI in a TIC was calculated as the mean SI value in the first three images. The maximum SI (SI max ) was defined as the peak enhancement value in the first pass of contrast. The contrast rise time (T rise ) was the time interval between SI baseline and SI max. The peak enhancement percentage [(SI max SI baseline )/SI base - line 100%] and relative enhancement slope [(SI max SI baseline )/SI baseline /T rise 100%] of each ROI were calculated and compared with other groups. An F-test was performed first to test the equality of variances, then an unpaired t-test was performed with equal or unequal variances according to the significance of the F-test between each groups. Statistical software Stata 7.0 (Stata Corp., TX) was used for calculation. Peak enhancement percentage SI max SI baseline SI baseline 100% Enhancement slope SI max SI baseline SI baseline T rise 100% Patterns of TIC The TICs were obtained by plotting the relative enhancement percentage [(SI c SI baseline )/SI baseline 100%] over time (0 80 seconds), where SI c is the postcontrast SI. The TICs were classified into five types (Fig. 3). Type A represents a nearly flat TIC. Type B represents a slow inclination curve. Type C represents a rapidly rising slope (wash-in) during the early phase, followed by a plateau after the peak enhancement is achieved. Type D represents a rapidly rising slope (wash-in) during the initial short period, the same as the upright portion of the type C curve, followed by a wash-out phase in the latter portion. Type E represents an initial rapidly rising slope followed by a second slowrising phase. If the difference of maximal enhancement

3 310 Chen et al. Figure 1. A 0-year-old male who had hepatocellular carcinoma for 1 year complained of lower back pain 2 weeks prior to imaging. a: An MR image of the lumbar spine (T1-weighted sagittal image (TR/TE 600/20)) showed hypointense lesions at the whole third lumbar vertebral body and the posterior half of the fifth lumbar vertebral body. b: T1-weighted sagittal image after Gd-DTPA administration with a fat suppression technique showed strong heterogeneous enhancement of the third and fifth lumbar vertebrae. c: Time-intensity graph derived from the low signal area of the fifth lumbar vertebra (shown in part a) showed a type D curve, with a wash-out phenomenon after peak enhancement. The y axis shows the SI in arbitrary units. and the SI at 80 seconds is greater than 20% of baseline SI, the C curve will be defined as type D or E, depending on the wash-in or wash-out of contrast. The cut-off value was set at 20% because it is easier to observe in a normalized TIC with a scaling format. RESULTS The average and standard deviations (SDs) for peak enhancement percentage in each group are summa- rized in Tables 1 and 2. As revealed in Table 1, acute compression fracture in group I subjects had a higher peak enhancement percentage compared to chronic compression fracture in group II subjects ( vs ; P 0.0). Metastatic vertebrae in group III subjects had a higher peak enhancement percentage as compared to chronic compression fracture in group II ( vs ; P 0.001). Pathologic fracture subjects in group IV had a higher peak enhancement percentage as compared to

4 Blood Perfusion of Vertebral Lesions 311 subjects had a higher enhancement slope as compared to chronic compression fracture in group II ( vs ; P 0.0). Pathologic fracture subjects in group IV had higher enhancement slope as compared to chronic compression fracture in group II (8.13. vs ; P 0.0). No significant difference for peak enhancement percentage and enhancement slope was noted in comparison with acute compression fracture and metastatic vertebrae with or without fracture, nonfractured metastatic vertebrae, and pathologic vertebral fracture. Figure 2. TIC of a dynamic Gd-DTPA-enhanced MR study. The various parameters used to derive the enhancement percentage and slope are shown. chronic compression fracture in group II ( vs ; P 0.0). The averages for normalized enhancement slope and their SD for the different groups are listed in Table 2. Acute compression fracture in group I subjects revealed a higher enhancement slope compared to chronic compression fracture in group II subjects ( vs ; P 0.0). Metastatic vertebrae in group III Patterns of TIC The data were reviewed by two radiologists on separate occasions, and consensus was obtained for the curve pattern classification. The distribution of the five TIC patterns in different groups is listed in Table 3. Furthermore, data from groups I and II were regarded as benign compression fracture groups, and data from groups III and IV were regarded as malignant groups. All 12 type D curves (100%) were found in malignant groups (groups III and IV) (Fig. 1). Six (8.7%) of seven type E curves (Fig. ) were revealed in benign compression fracture groups (groups I and II). Type C was the most common type in malignant groups (19/38, 0%), and the second-most common type in benign compression fracture (10/33, 30.3%). The type A curve was found in chronic compression fracture cases (/21, 19%). The type B curve was found in benign compression fracture (13/33, 36.3%) and nonfractured metastatic vertebrae (/32, 1.6%). Figure 3. Five types of TIC plotted from a dynamic MR study of the spine. Type D represents an early wash-out phenomenon. Type E represents a second slower-rising curve after a rapid contrast wash-in.

5 312 Chen et al. Table 1 Average and Standard Deviation for Peak Enhancement Percentage of Diseased Vertebral Bodies No. of patients No. of measured vertebral segments Enhancement percentage Group I Acute CF, symptom duration less than one month N 11 N Group II Chronic CF, symptom duration more than one month N 17 N Group III Metastatic vertebral lesion without pathologic fracture N 17 N Group IV Metastatic vertebral lesion with pathologic fracture N 6 N CF, compression fracture; SD, standard deviation. SD DISCUSSION Several factors contribute to the contrast enhancement of tissue (18): 1) a vascular supply, 2) an interstitial space for sequestering the contrast material, and 3) a route for the contrast material getting out of the vasculature. The rapid rise in SI immediately after the injection reflects the presence of gadopentetate dimeglumine in intravascular space as well as an increasing contribution of gadopentetate dimeglumine in extracellular space (16). An equilibration phase may be reached if there is a balance between the intra- and extravascular compartments, which represents the plateau of the type C curve. The type C curve was seen in most of the normal vertebrae, benign compression fractures, and malignant vertebral lesions (Table 3). Therefore, a type C curve is not helpful for differentiating between benign and malignant vertebral lesions. In a pilot study, we performed a contrast-enhanced dynamic pulse sequence in some cases for a longer observation time (280 seconds). The major event in the late phase was a gradual wash-out of contrast from tissue, which was observed in both compression-fracture and malignant groups, so we decided to focus on the first-pass and early-phase contrast enhancement and wash-out patterns. There were in total 13 type D curves (rapid contrast wash-in followed by early washout), presented in 71 measurements (Table 3). All of the type D curves were found in metastatic disease with or without associated pathologic fracture. We considered that the early washout phenomenon is specific for differentiating metastatic disease from benign compression fracture, because no type D curve occurred in the latter group. However, since the type D curve was found in only one-third (13/38, 3.2%) of malignant vertebral lesions, the sensitivity of the type D curve for diagnosing malignant vertebral lesions is not sufficiently high. The early wash-out pattern was previously described by Van Der Woude et al (7). This pattern was encountered in malignant bone tumors with high cellularity. The areas were histopathologically characterized by packed viable tumor cells with little intercellular matrix. A rapid contrast wash-in is followed by early wash-out because of the small volume of the interstitial space in areas consisting of packed viable tumor cells with a scarcity of matrix. Subsequently, SI decreases because of a lower concentration of gadopentetate demeglumine inflow. Early wash-out of contrast was also revealed in breast cancer studies by Daniel et al (19) and Kuhl et al (20) using Gd-DTPA-enhanced dynamic MRI, in which it was concluded that the washout phenomenon is specific for discriminating between benign and malignant disease. There were seven type E curves, presented in 71 measurements (Table 3). Most of the type E curves (6/7, 8.7%) were found in benign compression fracture cases (groups I and II). The fracture healing process has been arbitrarily divided into four phases: callus proliferation, vascularization, calcification, and reorganization. These phases overlap and can also be classified as an inflammatory phase, reparative phase, and a remodeling phase (21). An increase in vascularity at the fractured site is seen in the inflammatory phase, associated with infiltration of inflammatory cells, vasodilatation, and exudation of plasma. Due to an increase in vessel permeability in the inflammatory reaction, the type E Table 2 Average and Standard Deviation of Relative Enhancement Slope of Diseased Vertebral Bodies No. of patients No. of measured vertebral segments Average slope Group I Acute CF, symptom duration less than one month N 11 N Group II Chronic CF, symptom duration more than one month N 17 N Group III Metastatic vertebral lesion without pathologic fracture N 17 N Group IV Metastatic vertebral lesion with pathologic fracture N 6 N CF, compression fracture; SD, standard deviation. SD

6 Blood Perfusion of Vertebral Lesions 313 Table 3 Distribution of the Five Types of Normalized Time-Intensity Curves in Different Vertebral Lesions Group I. II. III. IV. acute, benign CF chronic, benign CF metastasis without fracture pathologic fracture Type A B C D E Total CF, compression fracture. TIC might be explained by more contrast sequesters into extracapillary space in the inflammatory phase without a balanced washout through venous drainage. A type E TIC was also revealed previously by Konig et al (22). By using a fast gradient-echo Gd-DTPA-enhanced dynamic MR pulse sequence, a second rising slope was shown in the hypervascular pannus of rheumatoid ar- thritis. The pannus is a highly inflammatory synovial tissue characterized by edema, infiltration of inflammatory cells, hypertrophy of synovial lining cells, and proliferative granulation tissues (23). We hypothesized that the cause of a second rising slope might be due to increased contrast sequestering into extracapillary space in an inflammatory event, and that type E is preferential for benign disease processes. However, since only six out of 33 (18.2%) vertebral compression fractures showed a type E curve, the sensitivity of the type E curve for diagnosing benign vertebral compression fractures is low. There was no significant difference for peak enhancement percentage and enhancement slope between benign acute compression fracture, metastatic vertebral lesion, and pathologic compression fracture. There were overlapping areas between these groups, which will cause diagnostic difficulty only when their peak enhancement percentage and slope are examined. According to Erlemann et al () and Verstraete et al (6), an Figure. A 7-year-old female complained of lowerback pain for 3 weeks prior to MR imaging study. a: An MR image of the thoracolumbar spine (T2-weighted sagittal image (TR/TE 2000/80)) with a fat suppression technique showed a fracture of the 12th thoracic vertebral body with bone marrow edema. b: T1-weighted sagittal image (TR/TE 600/20) after Gd-DTPA administration with a fat suppression technique showed marked heterogeneous enhancement of the fractured 12th thoracic vertebral body. c: Time-intensity graph derived from the 12th thoracic vertebra (as shown in part b) showed a type E curve, with a second slower-rising slope after a rapid contrast wash-in. The y axis shows the SI in arbitrary units.

7 31 Chen et al. overlap in the slope values was also demonstrated in the high vascular benign and malignant lesions. In the early inflammatory phase of fracture healing, an increase in vascularity 20 times greater than the resting circulation at the fractured site has been reported (2). Thus, high peak enhancement percentage and enhancement slope were observed in benign acute compression fracture. When conventional MR images for acute benign compression fracture cases were reviewed, the recently collapsed vertebral bodies all showed marked bone marrow edema in T2-weighted images, and the evidence of hyperemia was also observed from post-enhanced MRI (Fig. ). The proliferation of vascularity during fracture healing lasts about 3 10 weeks (21). In our study, the peak enhancement percentage and enhancement slope in acute compression fracture were significantly higher than that of the chronic compression fracture group, corresponding well to the fracture healing process. Several manifestations of MRI have been proposed as useful adjuncts in differentiating benign from malignant vertebral fractures (12 1). The findings that were preferential for a pathologic fracture were: complete marrow replacement, paravertebral soft-tissue lesion, pedicle involvement, abnormal expansile lesion with illdefined margin, and strong heterogeneous or irregular nodular-type soft-tissue infiltration. However, diagnostic difficulties may be encountered in some cases. The Gd-DTPA-enhanced dynamic MR study may provide additional information in equivocal cases. In conclusion, low enhancement percentage and shallow slope are more suggestive of a benign vertebral lesion. A type E curve (rapid contrast wash-in with a second rising slope) is more likely to occur in benign compression fracture. A type D curve (rapid contrast wash-in with early wash-out phenomenon) is preferential for metastatic disease of vertebra with or without associated pathologic fracture. Dynamic Gd-DTPA-enhanced MRI provides useful information for differentiating between benign and malignant vertebral lesions in certain cases. REFERENCES 1. Brach RC, Weinmann HJ, Wesbey GE. Contrast-enhanced NMR imaging: animal studies using gadolinium-dtpa complex. AJR Am J Roentgenol 198;12: Revel D, Brasch RC, Paajanen H, et al. Gd-DTPA contrast enhancement and tissue differentiation in MR imaging of experimental breast carcinoma. Radiology 1986;18: Pettersson H, Ackerman N, Kaude J, et al. Gadolinium-DTPA enhancement of experimental soft tissue carcinoma and hemorrhage in magnetic resonance imaging. Acta Radiol 1987;28: Erlemann R, Reiser MF, Peters PE, et al. Musculoskeletal neoplasm: static and dynamic Gd-DTPA-enhanced MR imaging. Radiology 1989;171: Fletcher BD, Hanna SL, Fairclough DJ, et al. Pediatric musculoskeletal tumors: use of dynamic, contast-enhanced MR imaging to monitor response to chemotherapy. Radiology 1992;18: Verstraete KL, De Deene Y, Roels H, et al. Benign and malignant musculoskeletal lesion: dynamic contrast-enhanced MR imagingparametric first-pass images depict tissue vascularization and perfusion. Radiology 199;192: Van Der Woude HJ, Bloem JL, Verstraete KL, et al. Osteosarcoma and Ewing s sarcoma after neoadjuvant chemotherapy: value of dynamic MR imaging in detecting viable tumor before surgery. AJR Am J Roentgenol 199;16: Bollow VM, Knauf W, Korfel A, et al. Initial experience with dynamic MR imaging in evaluation of normal bone marrow versus malignant bone marrow infiltrations in humans. J Magn Reson Imaging 1997; 7: Gold RI, Seeger LL, Bassett LW, et al. An integrated approach to the evaluation of metastatic bone disease. Radiol Clin North Am 1990; 28: Jacobsson H, Goransson H. Radiological detection of bone and bone marrow metastases. Med Oncol Tumor Pharmacother 1991; : Yamaguchi T, Tamai K, Yamato M, et al. Intertrabecular pattern of tumors metastatic to bone. Cancer 1996;78: Yuh WT, Zachar CK, Barloon TJ, et al. Vertebral compression fractures: distinction between benign and malignant causes with MR imaging. Radiology 1989;172: Baker LL, Goodman SB, Perkash I, et al. Benign versus pathologic compression fractures of vertebral bodies: assessment with conventional spin-echo, chemical-shift, and STIR MR imaging. Radiology 1990;17: Shih TT, Tsuang YH, Huang KM, et al. Magnetic resonance imaging of vertebral compression fractures. J Formos Med Assoc 1996;9: Shih TT, Huang KM, Li YW. Solitary vertebral collapse: distinction between benign and malignant causes using MR patterns. J Magn Reson Imaging 1999;9: Cova M, Kang YS, Tsukamoto H, et al. Bone marrow perfusion evaluated with gadolinium-enhanced dynamic fast MR imaging in a dog model. Radiology 1991;179: Stroszczynski C, Hosten N, Amthauer H, et al. Dynamic computerized tomography of the bone marrow of normal and pathologic vertebrae. ROFO Fortschr Geb Rontgenstr Nuklearmed 1997;167: Ross JS, Delamarter R, Hueftle MG, et al. Gadolinium-DTPA-enhanced MR imaging of the postoperative lumbar spine: time course and mechanism of enhancement. AJNR Am J Neuroradiol 1989; 10: Daniel BL, Glover GH, Ikeda DM, et al. Breast disease: dynamic spiral MR imaging. Radiology 1998;209: Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999;211: Cruess RL, Dumount J: Healing of bone, tendon and ligament. In: Rockwood Jr CA, Green DP, editors. Fractures. Philadelphia: JB Lippincott; 197. p Konig H, Sieper J, Wolf K. Rheumatoid arthritis: evaluation of hypervascular and fibrous pannus with dynamic MR imaging enhanced with Gd-DTPA. Radiology 1990;176: Resnick D, Niwayama G. Rheumatoid arthritis and the seronegative spondyloarthropathies: radiographic and pathologic concepts. In: Resnick D, Niwayama G, editors. Diagnosis of bone and joint disorders. 3rd ed. Philadelphia: Saunders; p Rheinlander FW. The normal micro-circulation of diaphyseal cortex and its response to fracture. J Bone Joint Surg 1968;0: