European Radiology

, Volume 13, Issue 8, pp 1849–1858

Dynamic contrast-enhanced MR imaging in monitoring response to isolated limb perfusion in high-grade soft tissue sarcoma: initial results

Authors

    • Department of RadiologyLeiden University Medical Center
  • Maartje J. A. Geirnaerdt
    • Department of RadiologyNetherlands Cancer Institute Antoni van Leeuwenhoek Hospital
  • Pancras C. W. Hogendoorn
    • Department of PathologyLeiden University Medical Center
  • Johannes L. Peterse
    • Department of PathologyNetherlands Cancer Institute, Antoni van Leeuwenhoek Hospital
  • Frits van Coevorden
    • Department of SurgeryNetherlands Cancer Institute, Antoni van Leeuwenhoek Hospital
  • Antonie H. M. Taminiau
    • Department of Orthopedic SurgeryLeiden University Medical Center
  • Rob A. E. M. Tollenaar
    • Department of SurgeryLeiden University Medical Center
  • Bin B. R. Kroon
    • Department of SurgeryNetherlands Cancer Institute, Antoni van Leeuwenhoek Hospital
  • Johan L. Bloem
    • Department of RadiologyLeiden University Medical Center
Musculoskeletal

DOI: 10.1007/s00330-002-1785-4

Cite this article as:
van Rijswijk, C.S.P., Geirnaerdt, M.J.A., Hogendoorn, P.C.W. et al. Eur Radiol (2003) 13: 1849. doi:10.1007/s00330-002-1785-4

Abstract

The objective of this study was to evaluate whether dynamic contrast-enhanced MR imaging can determine tumor response and localize residual viable tumor after isolated limb perfusion (ILP) chemotherapy in soft tissue tumors. Twelve consecutive patients, with histologically proven high-grade soft tissue sarcoma, prospectively underwent non-enhanced MR and dynamic contrast-enhanced MR imaging before and after ILP. Tumor volume was measured on non-enhanced MR images. The temporal change of signal intensity in a region of interest on dynamic contrast-enhanced MR images was plotted against time. Start, pattern, and progression of enhancement were recorded. Histopathologic response was defined as complete response if no residual viable tumor was present, partial remission if <50% viable tumor was present, and no change if ≥50% viable tumor was present in the resection specimen. Resected specimens for correlation with histopathology were available for 10 patients; 5 patients had partial remission and 5 had no change. Volume measurements correctly predicted tumor response in 6 of 10 patients. Dynamic contrast-enhanced MR correctly predicted tumor response in 8 of 10 patients. Early rapidly progressive enhancement correlated histologically with residual viable tumor. Late and gradual, or absence of enhancement, was associated with necrosis, predominantly centrally located, or granulation tissue. These preliminary results show that dynamic contrast-enhanced MR imaging offers potential for non-invasive monitoring of response to isolated limb perfusion in soft tissue sarcomas due to identification of residual areas of viable tumor and subsequently may provide clinically useful information with regards to timing and planning of additional surgery. Further prospective studies in a larger patient population is warranted.

Keywords

Soft tissueNeoplasmsMagnetic resonanceDynamic contrast enhancementTherapy monitoringIsolated limb perfusion

Introduction

Anti-angiogenesis and anti-neovasculature therapy has been proposed in various types of cancer [1, 2]. Anti-angiogenesis therapy involves inhibiting the process of new vessel formation to prevent any further growth of primary tumor and metastases, whereas anti-neovascular therapy attacks the existing neovasculature. Recombinant tumor necrosis factor α (TNF) is an anti-neovascular agent, usually administrated concomitantly with chemotherapeutic agents, such as melphalan, in high doses by regional isolated limb perfusion (ILP). By inducing regression of the primary tumor, ILP with TNF and melphalan may allow local surgery with limb preservation instead of amputation or other functionally mutilating surgery in patients with locally advanced and/or high-grade soft tissue sarcoma [3, 4, 5]. The ILP with TNF and melphalan induces exclusive destruction of the tumor vasculature and hemorrhagic tumor necrosis [6, 7]. These selective destructive effects on tumor-associated vessels have been evaluated by angiography [8]. As shown with angiography, rapid disappearance of tumor-associated vascularity is indicative for tumor response, whereas persistent pathologic vascularity is associated with progressive disease [8, 9]. A non-invasive method allowing identification of viable tumor, in relation to anatomical landmarks, based on intra-tumoral perfusion might be clinically useful in timing and planning of additional surgery. In particular, identification of patients demonstrating progressive disease is of vital importance to prevent delay of treatment.

Assessment of tumor blood supply and perfusion has become feasible with the use of fast dynamic contrast-enhanced MR imaging. Dynamic MR imaging with high temporal resolution permits non-invasive evaluation of the distribution of gadopentetate dimeglumine (Gd-DTPA) into the tumor capillaries and its subsequent diffusion through the pathological permeable capillaries into the interstitial space [10]. This technique has been applied to previous clinical studies as a potential early predictor of tumor response in osteosarcoma and Ewing sarcoma [11, 12, 13, 14, 15, 16]. The purpose of our study was to evaluate the accuracy of dynamic contrast-enhanced MR imaging in determining tumor response to ILP, and in identifying residual areas of viable tumor in patients with high-grade soft tissue sarcoma.

Materials and methods

Patients

Between July 1998 and October 2000, 12 consecutive patients (2 women, 10 men; age range 15–79 years, median age 54 years) with histologically proven high-grade soft tissue sarcoma, in which initial local resection was not possible, were prospectively included. Informed consent was obtained from all patients. The primary objective of ILP in all patients was to avoid amputation or major mutilating surgery. Two patients were excluded from analysis: 1 patient developed pulmonary metastases after ILP and was thus not a candidate for surgery, and the other patient had an early amputation 2 days after commencing ILP because of necrosis of the hand secondary to ILP. Histologic diagnoses were established by correlating light microscopic features with the immunocytochemical profile and included monophasic (n=2) and biphasic synovial sarcoma (n=2), high-grade pleiomorphic sarcoma not otherwise specified (NOS; n=2), pleiomorphic (n=1) and myxoid/round cell liposarcoma (n=1), and one each with a pleiomorphic rhabdomyosarcoma and a leiomyosarcoma, respectively (Table 1).
Table 1.

Patient characteristics, MR imaging findings, and pathologic response. ILP isolated limb perfusion, CR complete remission, PR partial remission, NC no change, PD progressive disease, n.a. not applicable, NOS not otherwise specified

Patient characteristics

MR imaging findings

Histopathologic examination

No.

Gender, age (years)

Location

Tumor volume ratioa

Standard non-enhanced MR response

Areas of early enhancement after ILP

Pattern of early enhancement after ILP

Progression of early enhancement after ILP

Dynamic MR response

Tumor type

Macro/microscopic view

Pathologic response

1

M, 79

Lower arm

1.01

NC

Present

Multi-focal

Type III

NC

High-grade pleomorphic sarcoma (NOS)

Viable tumor

NC

2

M, 56

Upper leg

2.28

PD

Present

Peripheral rim

Type V

PR

Pleiomorphic liposarcoma

Central necrosis, minute foci of viable tumor

PR

3

M, 66

Lower leg

1.16

NC

Present

Multi-focal

Type V

NC

Monophasic synoviosarcoma

Viable tumor

NC

4

M, 20

Upper arm

0.89

NC

Present

Multi-focal

Type V

NC

High-grade pleomorphic sarcoma NOS

Viable tumor

NC

5

M, 51

Lower leg

0.78

NC

Present

Focal areas and a predominant peripheral rim

Type V

PR

Pleiomorphic Rhabdomyosarcoma

Extensive necrosis with areas of viable tumor

PR

6

M, 41

Lower leg

0.19

PR

Absent

n.a.

n.a.

CR

Myxoid/round cell liposarcoma

Well-differentiated low-grade areas mixed with areas of high-grade myxoid liposarcoma

PR

7

M, 31

Lower arm

2.10

PD

Present

Multi-focal

Type III

NC

Biphasic synoviosarcoma

Viable tumor

NC

8

F, 56

Upper leg

0.60

PR

Absent

n.a.

n.a.

CR

Monophasic synoviosarcoma

Minute foci of viable tumor

PR

9

F, 15

Lower arm

0.36

PR

Present

Multi-focal

Type V

NC

Biphasic synoviosarcoma

Viable tumor

NC

10

M, 70

Foot

0.85

NC

Present

Peripheral rim

Type V

PR

Leiomyosarcoma

Central necrosis, viable tumor adjacent to vascular walls peripherally

PR

aTumor volume ratio expresses the change in tumor volume by dividing the tumor volume after ILP by the initial tumor volume

The perfusion technique used is based on the technique developed by Creech et al. [17]. The cannulated arteries and veins of the perfused extremity are connected to an extracorporeal circuit. A tourniquet is placed at the base of the extremity to minimize systemic leakage. The extremity is arterially perfused with TNF and followed later by melphalan.

MR imaging

Patients were scheduled for two MR examinations: one before ILP (range 1–92 days, median 20 days before perfusion) and one after ILP before tumor resection (range 35–150 days, median 56 days after perfusion). The median interval between the second MR examination (performed after ILP) and tumor resection was 27 days (range 1–53 days).

The MR imaging was performed either on a Philips 1.5-T NT15 MR system (Philips Medical Systems, Shelton, Conn.) in 7 patients or on a Siemens 1.5-T Magnetom Vision MR system (Siemens Medical Systems, Erlangen, Germany) in 3 patients, using similar pulse sequences. The same MR system was used for each individual patient and in all patients a surface coil was used. Standard non-enhanced T1- and T2-weighted fast-spin-echo sequences were followed by a fast dynamic contrast-enhanced MR sequence. On both MR systems we used for dynamic imaging a T1-weighted gradient-echo sequence with a temporal resolution of 3 s. We used a turbo-field-echo (TFE) sequence on the Philips system: TR/TE=5.4 ms/1.4 ms; flip angle 20°; preparatory pulse delay time 165 ms; one excitation per dataline; matrix size 102×256; field of view 300–400 mm; section thickness 5–8 mm; number of sections 8. We used a fast low-angle shot (FLASH) 2D sequence on Siemens: TR/TE=29.2 ms/1.4 ms; flip angle 30°; one excitation; matrix size 95×128; field of view 400 mm; section thickness 8–10 mm; and number of sections 7. The intravenous bolus injection of 0.1 mmol/kg body weight Gd-DTPA (Magnevist, Schering, Berlin, Germany) was started 5 s after start of data acquisition. The injection rate, using a power injector, was 2 ml/s, immediately followed by a saline flush of 20 ml at the same injection rate. The total scan time was 5 min. Section orientations were selected in the longitudinal plane, best exhibiting the tumor and an artery. The second pre-contrast dynamic image was subtracted from the dynamic contrast-enhanced MR images by using standard commercially available software. Arbitrary shaped regions of interest were drawn on the dynamic subtraction images in the early enhancing areas of the tumor and in non-enhancing areas within the tumor. The signal intensity values during the dynamic study were plotted against time as time—intensity curves. We subjectively classified five different time—intensity curves: absence of enhancement (type I); gradual increase of enhancement (type II); rapid initial enhancement followed by a plateau phase (type III); rapid initial enhancement followed by a washout phase (type IV); or rapid initial enhancement and sustained late enhancement (type V) [18]. T1-weighted spin-echo sequences were repeated in two perpendicular planes approximately 10 min after injection of Gd-DTPA.

Analysis

On standard T2-weighted spin-echo MR images we determined tumor volume before and after isolated limb perfusion. Tumor volume measurements were performed by manually outlining the tumor margins on each image of the T2-weighted series by using standard software on an EasyVision workstation (Philips Medical Systems, Shelton, Conn.). Change in mean tumor volume was expressed as a ratio by dividing the tumor volume after ILP by the initial tumor volume. Standard non-enhanced MR response based on these tumor volume ratios was correlated with histologic response. We considered a decrease in tumor volume of at least 35% to represent a partial response (tumor volume ratio 0.65 or lower). Accordingly, tumor volume ratios between 0.65 and 1.35, and higher than 1.35 were interpreted, respectively, as no change and progressive disease [19]. Difference in tumor volume ratio between the histologic response categories was evaluated using Student's t-test.

Two radiologists independently evaluated all dynamic MR images without knowledge of the histologic response or patient outcomes. Both observers identified on dynamic subtraction images areas within the tumor that demonstrated early (within 6 s after arrival of the contrast bolus in an artery) and rapidly progressive enhancement. These areas were considered to represent viable tumor. Subsequently, areas of late (more than 6 s after arrival of the contrast bolus in an artery) and gradual or absence of enhancement were considered to represent non-viable tumor. Based on results with the first pass of Gd-DTPA after injection of 2 ml/s in extremity musculoskeletal tumors, this arbitrary threshold of 6 s (interval arterial-, lesional enhancement) was chosen [20, 21, 22]. The subjective assessment was verified by producing time—intensity curves. Areas of early and rapidly progressive enhancement were verified by type-III, type-IV, or type-V time—intensity curves. Subsequently, areas within the tumor demonstrating late and gradual or absence of enhancement were verified by type-I or type-II time—intensity curves. Finally, the findings on dynamic contrast-enhanced MR imaging before and after ILP were compared, and changes in the amount of viable tumor were classified. Dynamic MR response was defined as complete response if no early enhancement was present after ILP indicating absence of residual viable tumor, partial remission if <50% of the remnant tumor mass demonstrated early enhancement, and no change if ≥50% of the remnant tumor mass demonstrated viable tumor.

Histopathologic response was determined by examination of the resected residual tumor specimens. At resection surgical specimens were oriented with sutures for comparative histopathologic study. Macroslabs were obtained from all specimens in a plane identical to that of the dynamic contrast-enhanced MR images. These macroslabs were photographed and representative tissue samples were taken, encompassing macroscopically different tumor areas, including necrosis. Based on an integration of gross and microscopic findings, a semi-quantitative estimation of the amount of remnant viable and necrotic or regressive tumor was made. Histopathologic response was defined as complete response if no residual viable tumor was present in the serial sectioned specimen after complete surgical resection, partial remission if <50% viable tumor was present of the remnant tumor volume, and no change if ≥50% viable tumor was present of the remnant tumor volume. Subsequently, the histologic macroslabs were compared in detail to the corresponding dynamic contrast-enhanced MR images by a radiologist and a pathologist. The presence and localization of residual viable tumor foci was correlated with dynamic enhancement.

Results

Histologic response after ILP

Histologic response and MR imaging results for each patient are summarized in Table 1. On histopathologic examination, 5 patients had partial remission and the other 5 had no change. None of the patients had complete response.

Tumor volume and histopathologic classification of response

No significant difference in tumor volume ratio was observed between the histologic response categories (p>0.05). Mean tumor volume ratio was 1.10 and 0.93 in, respectively, the histologically unchanged and the partial remission group (Table 2). In 3 of 10 patients tumor volume decreased (at least 35%) after ILP, on histopathologic examination 2 patients had partial remission, and 1 patient had no change. In 2 patients tumor volume increased (at least 35%), on histological examination 1 patient had partial remission (Fig. 1), and 1 patient had no change. Five patients demonstrated stable tumor volume after ILP, with histologic partial remission in 2 patients and no change in 3 patients; hence, standard MR measurements of tumor volume response resulted in a correct prediction of tumor response in 6 of 10 patients (three false negatives and one false positive; Table 2).
Table 2.

Histologic response to ILP vs standard MR response based on volume measurements

Histologic response

Histologic response

PR

NC

Absence of tumor (CR)

0

0

Decreased tumor volume (PR)a

2

1

Stable tumor volume (NC)b

2

3

Increased tumor volume (PD)

1

1

Mean tumor volume ratio

0.93

1.10

aDecrease in tumor volume of at least 35% (tumor volume ratio 0.65 or lower) was considered to represent a partial response

bTumor volume ratios between 0.65 and 1.35 were interpreted as NC

Fig. 1a–g.

Pleiomorphic liposarcoma in the upper leg of a 56-year-old man. a Sagittal T1-weighted (TR 550 ms, TE 7 ms) MR image performed before isolated limb perfusion (ILP) displays a soft tissue mass in close contact to the neurovascular bundle (arrows) in hamstring compartment. b Corresponding sagittal T1-weighted (TR 550 ms, TE 7 ms) MR image performed 6 weeks after ILP. Tumor volume is markedly increased. c Selection of five consecutive sagittal dynamic contrast-enhanced MR images acquired before ILP demonstrates irregular early tumor enhancement (arrow), within 6 s after arrival of the bolus contrast agent in the artery (arrowhead), indicating viable tumor. d Selection of five consecutive sagittal dynamic contrast-enhanced MR images acquired after ILP demonstrates a peripheral rim of early enhancement (arrow). The central part of the tumor does not enhance. Dynamic MR response was determined as partial remission. e Photograph of cut section of surgical specimen in the identical plane of the dynamic sequence after ILP with TNF and melphalan shows the well-demarcated tumor with central necrosis (arrow). f Microscopy of central part demonstrating vital tumor component of scattered pleomorphic atypical lipoblast-like cells (hematoxylin and eosin, original magnification ×200). These minute foci of viable tumor embedded in extensive areas of tumor necrosis were not depicted on dynamic imaging. g Microscopy of peripheral part of the tumor (hematoxylin and eosin, original magnification ×200) demonstrating the tumor capsule. Histopathologic examination of the surgical specimen identified a partial remission

Dynamic MR and histopathologic classification of response

The initial dynamic contrast-enhanced MR studies obtained before Tru-cut biopsy and ILP showed extended areas of early and rapidly progressive enhancement in all patients. After ILP, 8 of 10 patients showed residual areas of early enhancement on dynamic MR imaging. These areas were encountered either peripherally or randomly distributed as multi-focal solid tumor fields through the remnant tumor mass. In 5 patients multi-focal early rapidly progressive enhancement displaying type-III, type-IV, or type-V time—intensity curves was observed in ≥50% of the remnant tumor mass. Histopathology demonstrated no change in these 5 patients (Fig. 2). In 3 patients predominantly peripheral early and rapidly progressive enhancement displaying type-III, type-IV, or type-V time—intensity curves was observed in <50% of the remnant tumor mass. All 3 patients had histopathologic partial remission; however, in 1 of these 3 patients there was an early enhancing peripheral rim without viable tumor histologically but also isolated scattered foci of viable tumor cells seen on microscopy not depicted on dynamic contrast-enhanced MR imaging that resulted in histopathologic partial remission. Two patients were classified on dynamic MR as complete response because no early dynamic enhancement was demonstrated after ILP indicating absence of residual viable tumor. One of these 2 patients had only isolated scattered foci of viable tumor cells, seen on microscopy (histopathologic partial remission). The other patient had larger areas of viable tumor of variable morphology not depicted by dynamic contrast-enhanced MR imaging. Dynamic contrast-enhanced MR imaging correctly predicted response in 5 patients with histologic no change, and in 3 patients with histologic partial remission. Two patients were incorrectly predicted to have complete response, whereas histologic evaluation showed partial remission (patients 6 and 8; Table 1).
Fig. 2a–e.

Synovial sarcoma in the lower arm of a 31-year-old man. a Transverse T2-weighted (TR 3694 ms, TE 60 ms) MR image with fat suppression acquired before ILP demonstrates a homogeneous high signal intensity soft tissue mass located in the flexor compartment of the lower arm. b Transverse T2-weighted (TR 2973 ms, TE 60 ms) MR image with fat suppression following ILP. Tumor dimensions were not decreased. c Selection of four consecutive sagittal dynamic contrast-enhanced MR images before ILP. The arrow marks the start of arterial enhancement with immediate multi-focal tumor enhancement. d Selection of four consecutive sagittal dynamic contrast-enhanced MR images after ILP. Multi-focal early (within 6 s after arterial enhancement) enhancement, indicating viable tumor, was observed in ≥50% of the remnant tumor mass. e Microscopy of early enhancing tumor exhibits residual viable tumor after ILP. The tumor is composed of densely packed cells with spindle-shaped nuclei characteristic of synovial sarcoma. No necrosis is observed (hematoxylin and eosin, original magnification ×200). Histopathologic response was recorded as not changed

Discordance between tumor volume response and dynamic MR response

Discordance between the dynamic MR response and tumor volume occurred in 6 patients. Based on tumor volume measurements, response was correctly classified in 2 patients as partial remission, whereas dynamic MR response indicated complete response (patients 6 and 8; Table 1). Tumor volume measurements incorrectly classified response in 4 patients, whereas dynamic MR classification was correct (patients 2, 5, 9, and 10; Table 1). One of these 4 patients was classified as partial remission based on volume measurement; however, on histopathologic examination response was not changed (patient 9; Table 1). The other 3 patients were classified as no change (2 patients) or progressive disease (1 patient) based on tumor volume; on histopathologic examination all 3 patients had partial remission (patients 2, 5, and 10; Table 1).

Detailed correlation between dynamic MR findings and histopathology

On histopathologic examination after ILP, multi-focal early enhancement corresponded to predominantly viable tumor tissue with only occasional small foci of sclerosis, hyalinization, or necrosis. Early enhancement of a peripheral rim, with variable thickness, was encountered in 3 patients: 2 of these patients demonstrated remaining viable tumor peripherally adjacent to vascular walls, and 1 patient demonstrated on microscopy a capsule containing substantial neovascularization and granulation tissue without viable tumor. On histopathologic examination 2 patients presented microscopic isolated minute foci of viable tumor cells that could not be visualized on the dynamic contrast-enhanced MR images (patients 2 and 8; Table 1).

Late and gradual, or absence of dynamic enhancement, was associated with hemorrhagic necrosis (tumor- or chemotherapy related), granulation tissue, or fibrous tissue. In general, the necrotic areas were centrally located. In 1 patient with a myxoid/round cell liposarcoma, only late and heterogeneous enhancing areas were found mixed between areas of absent dynamic enhancement. These heterogeneous late-enhancing areas corresponded to poorly vascularized areas of myxoid liposarcoma. The non-enhancing areas of the residual soft tissue mass were composed of mature lipocytes (patient 6; Table 1).

Discussion

Isolated limb perfusion has been advocated as being advantageous because some high-grade and/or locally advanced soft tissue sarcomas may become amenable to local surgical resection [3, 4, 5]. After ILP, early assessment of tumor response and localization of viable tumor can prove beneficial: patients exhibiting a favorable response can be treated by local surgical resection with preservation of the extremity, whereas poor responders should, if possible, be treated more aggressively to prevent poor outcome.

Up until now, clinical and radiologic volume measurements on standard non-enhanced MR images have been routinely used to measure tumor response to ILP preoperatively [4, 23]. According to previous reports and our own results, tumor volume is not a reliable predictor of tumor response to preoperative (systemic or local) chemotherapy in soft tissue sarcoma [24, 25, 26, 27, 28]. In this study we documented a 40% discrepancy between tumor volume response and histopathologic response. During ILP soft tissue sarcomas usually do not reduce in size dramatically, but may become softer, hemorrhagic, and edematous. Eventually, they may become predominantly necrotic without a definite change in tumor volume. After preoperative chemotherapy (local or systemic) an increase of tumor volume can be caused either by increase of viable tumor or increase of tumor necrosis and/or hemorrhage associated with a decrease of viable tumor [7, 25, 26, 27, 29]. Monitoring chemotherapy and planning surgery requires an imaging modality that is capable not only of depicting changes in tumor volume, but also of identifying residual viable tumor.

Tumor volume is also one of the problems in histopathologic assessment of tumor response on preoperative (systemic or local) chemotherapy in soft tissue sarcomas. Tumor volume changes cannot be taken into account in histologic analysis of response; therefore, it may happen that despite a considerable decrease in tumor volume, histopathologic response is classified as not changing because more than 50% of the remnant tumor volume is viable (patient 9; Table 1). Other limitations of histopathologic assessment of tumor response are that the complete surgical specimens cannot be analyzed. Moreover, histopathologic examination of surgical specimens after ILP reveals varying degrees of tumor destruction such as intratumoral hemorrhage and necrosis. The extent to which these changes represent treatment effect or the natural history of the tumor remains undetermined. Finally, histologic evaluation of response prior to resection is unreliable, because a preoperative histologic biopsy often does not reflect the amount of necrosis in the entire soft tissue mass, and is often complicated by the limited amount of biopsy material.

In our study, dynamic contrast-enhanced MR imaging allowed good correlation with the current histologic standard of percentage viable tumor vs tumor necrosis in 7 of 10 patients with high-grade soft tissue sarcoma before surgical resection. In 2 patients only microscopic residual minute foci of viable tumor embedded in extensive areas of tumor necrosis were not depicted on dynamic imaging (patients 2 and 8; Table 1). The limited spatial resolution of the dynamic sequence may account for these false-negative findings. Moreover, in 1 patient (patient 6; Table 1) with histopathologic partial remission, dynamic contrast-enhanced MR imaging failed to identify areas of viable tumor. Microscopically, these areas showed a variable morphology consisting of a mixture of highly differentiated lipocytic cells in a myxoid background with poor vascularization. Round cell areas were lacking.

Dynamic contrast-enhanced MR imaging visualizes the first pass of Gd-DTPA in tumor tissue and is a function of microvascular density and permeability [10, 30]. Absence or a decrease in areas of early enhancement present in 5 of 10 patients after ILP indicates that a large portion of the tumor is either necrotic, due to destruction of endothelial cells or vascular collapse secondary to high tumor interstitial pressure, or is still viable with reduced vascularization that could indicate decreased angiogenesis. These viable regions may have responded to ILP but are not necrotic. Patients with a less favorable response to ILP, present in 5 of 10 patients, demonstrate multi-focal regions of viable tumor consisting of ≥50% of the remnant tumor mass that are highly angiogenic. Persisting early enhancement after ILP indicates persisting tumor neovascularization, associated with poor response [8, 9, 31, 32, 33]. Subsequently, fast dynamic contrast-enhanced MR imaging may, especially in case of ineffective ILP, influence the timing and extensiveness of surgical intervention since further waiting can be life-threatening. Selection of standard time points for monitoring therapy is therefore recommended.

Several methods, such as angiography, positron emission tomography (PET), and MR spectroscopy, have been successfully used to predict response to ILP in patients with soft tissue sarcoma [8, 28, 34, 35, 36, 37, 38]. The advantage, however, of MR imaging, including dynamic contrast-enhanced MR imaging, over the above-mentioned imaging modalities is the ability to identify and localize areas of viable tumor after neoadjuvant chemotherapy/ILP, and to display anatomic relations and extent that is used to plan surgery [32, 33].

Spatial resolution in MR imaging has its limits, as dynamic contrast-enhanced MR images did not depict minute foci of residual tumor; however, both the absence of MR detectable residual viable tumor and the presence of only minute foci of residual viable tumor at histopathologic examination are predictive of improved local control. Adequate tumor sampling with multiple dynamic MR sections, high spatial resolution, and high temporal resolution are competitive conditions of dynamic contrast-enhanced MR imaging. High spatial resolution imaging with sufficient high temporal resolution and adequate tumor sampling may improve the detection of small foci of viable tumor [39]. In addition to these improvements at the level of data acquisition, improvements in image post-processing also have the potential to enhance MR performance. Pharmacokinetic modeling of data in each voxel of dynamic contrast-enhanced MR images has, for instance, been used successfully to estimate the relative volume of residual viable tumor in Ewing's sarcoma after neoadjuvant chemotherapy [40]. Other limitations of our study, besides the limited spatial resolution, are the fairly small number of included patients with only partial responders and non-responders caused by the low incidence of ILP representing a therapy modality in locally advanced soft tissue sarcomas. Another disadvantage is the relatively wide range between ILP and the first MR examinations, and between the second MR examination and ILP.

Conclusion

In conclusion, dynamic contrast-enhanced MR imaging can depict remnant areas of viable tumor and therefore may offer the potential for assessment of response to ILP (and systemic chemotherapy) in soft tissue sarcomas. Compared with tumor volume measurements on standard non-enhanced MR, it seems to be a more accurate tool in determining response. Compared with histopathologic examination, it is noninvasive, can sample the entire tumor volume, and may provide additional anatomic and physiologic information with regard to timing and planning of surgical intervention. These preliminary results warrant a further prospective study in a larger patient population.

Copyright information

© Springer-Verlag 2003