Use of intraoperative ultrasound for localizing tumors and determining the extent of resection: a comparative study with magnetic resonance imaging

Transcription

1 J Neurosurg 84: , 1996 Use of intraoperative ultrasound for localizing tumors and determining the extent of resection: a comparative study with magnetic resonance imaging MAAROUF A. HAMMOUD, M.D., B. LEE LIGON, PH.D., RABIH ELSOUKI, M.P.H., PH.D., WEI MING SHI, M.D., DONALD F. SCHOMER, M.D., AND RAYMOND SAWAYA, M.D. Departments of Neurosurgery and Radiology, The University of Texas M.D. Anderson Cancer Center, and The University of Texas School of Public Health, Houston, Texas A prospective study of 70 patients with intraparenchymal brain lesions (36 gliomas and 34 metastases) was performed to evaluate the efficacy of intraoperative ultrasound (IOUS) in localizing and defining the borders of tumors and in assessing the extent of their resection. Eighteen of the 36 glioma patients had no previous therapy. All of these 18 tumors were well localized by IOUS; margins were well defined in 15 and moderately defined in three. The extent of resection was well defined on IOUS in all 18 patients, as confirmed by measurements taken on postoperative magnetic resonance (MR) images (p = 0.90). The remaining 18 patients with gliomas had undergone previous surgery and/or radiation therapy; five had recurrent tumors and 13 had radiation-induced changes. The extent of resection of the recurrent tumors was well defined in all but one patient, as confirmed by postoperative MR imaging. The extent of resection was poorly defined in all 13 patients whose pathology showed radiation effects. All 34 metastatic lesions were well localized and had well-defined margins. In addition, IOUS accurately determined the extent of resection in all cases; the results were confirmed with postoperative MR imaging. In conclusion, IOUS is not only helpful in localizing and defining the margins of gliomas and metastatic brain lesions, it also accurately determines the extent of resection, as confirmed by postoperative MR imaging. This assessment does not apply, however, when the lesion is due primarily to radiation effect. KEY WORDS brain metastasis glioma magnetic resonance imaging radiation-induced lesion ultrasound B RAIN tumors are often rapidly fatal, particularly when they are malignant. The reported incidence of primary and metastatic brain tumors combined is 16.7 per 100,000 persons. 22 Uncontrolled and retrospective studies of surgical series have confirmed that the most effective management of malignant intracranial tumors, whether primary or metastatic, is surgical resection followed by adjuvant irradiation and chemotherapy. 2,15,19,23 25,28,31 33,35 Maximum surgical resection of brain tumors requires accurate localization and precise delineation of its margins and of any infiltration outside the margins. Preoperative computerized tomography (CT) and magnetic resonance (MR) imaging readily identify the morphological features of tumors, but neither one sufficiently depicts the margins of a solid tumor, infiltrating tumor cells, peritumoral edema, or normal brain adjacent to tumor. 10,11,20,21 Intraoperative ultrasound (IOUS) has been shown to have especially important potential because it identifies most brain lesions, 6,7,35 and differentiates solid tissue from liquefaction or cyst. 19 Its particular relevance for neurosurgery involves both its ability to provide the surgeon with immediate feedback and its help in achieving maximum resection of the tumor. 8,13,16,23,26,34 Despite its benefits, which have been reported since 1980, 29 there have been no studies evaluating the efficacy of IOUS in assessing the extent of resection of brain tumors as compared to postoperative MR images. Clinical Material and Methods We conducted a prospective study of 70 patients with intraparenchymal brain lesions (36 gliomas and 34 metastatic tumors) to evaluate the efficacy of IOUS in 1) localizing the tumors; 2) defining their margins; and 3) determining the extent of resections. There were 38 men and 32 women, who ranged in age from 20 to 90 years (median 49 years). Patients who had undergone previous surgery for tumor resection and/or radiation therapy and chemotherapy were noted. All patients underwent preand postoperative MR imaging with intravenous administration of gadolinium contrast agent. Both procedures were performed within 5 days of the operation. The MR images were scanned, digitized, and sorted in the comput- 737

2 M. A. Hammoud, et al. FIG. 3. Intraoperative ultrasound image of radiation-induced changes (RC) showing the poorly defined margins of the lesion. FIG. 1. Magnetic resonance image (left) and intraoperative ultrasound image (right) of juvenile pilocytic astrocytoma (T) showing the cyst (C), tumor margins (B), falx (F), edema (E), and anterior horns of the lateral ventricles (V). er, using commercially available software (Macintosh, Image 1.52). Gross solid tumor margins were arbitrarily delineated at the perimeter of signal change on T 1 -weighted contrast-enhanced images for enhancing and T 2 - weighted images for nonenhancing tumors. In defining any residual tumor, careful comparison was made between the pre- and postoperative MR images. During surgery, all 70 tumors were evaluated by one of the authors (R.S.) using a real-time ultrasound scanner (model OR340; Aloka, Wallingford, CT). The transducer frequencies included 5.0 and 7.5 MHz to maximize tumor and boundary resolutions. The entire operative field was scanned in a systematic fashion in at least two perpendicular planes before and after tumor resection. Tumors were found to be hyperechoic in comparison with surrounding edema or brain tissue (Fig. 1). During tumor removal, IOUS was used to guide the resection toward the echogenic margin in cases for which complete excision was the aim. After the dura was closed, ultrasound was used again to assess the extent of resection, and photographs of these images were taken. All ultrasound images obtained during FIG. 2. Intraoperative ultrasound images of left occipital metastatic melanoma (T) in coronal (left) and axial (right) views showing the well-defined margins of tumor (B), falx (F), and edema (E). surgery and after closure of the dura were digitized and stored in the computer. Localization was defined as either: well localized (location of tumor was well visualized by IOUS) or poorly localized (location of the tumor was not well visualized by IOUS). The margins were considered well defined when they could be clearly visualized and separated from surrounding edema and brain tissue (Fig. 2); moderately defined when they could be visualized by IOUS but could not be clearly separated from surrounding tissue in certain areas; and poorly defined when they could not be visualized or separated from surrounding tissue (Fig. 3). The extent of tumor resection was considered well defined when no residual tumor was seen because of total excision or residual tumor was clearly seen and further excision was not advisable; or poorly defined when the IOUS image did not allow a determination of whether the excision was complete. The efficacy of IOUS in defining the extent of resection was determined by comparing the volumes measured by IOUS images obtained at the end of the procedure with the volumes measured by postoperative MR images. Tumor volumes were estimated for both modalities using a geometric formula (prolate ellipsoid) based on maximum sagittal, coronal, and anteroposterior diameters: V = /6 A B C. 18 The volume of a prolate ellipsoid more accurately accounts for the complex shape of the tumors than do models based on spherical or cubic geometry. To compensate for any potential random error occasioned by variations in tumor morphology, we used the same observers (W.M.S., D.F.S.) and equipment and averaged several measurements. The estimates of tumor volumes by both MR and ultrasound images could, thus, be compared. Statistical Analysis The data for the entire series were expressed as the mean standard error of the mean. As in a study by Altman and Bland, 1 statistical comparison between different imaging modalities was performed by comparing mean differences to zero using a paired t-test. The underlying assumption of independence was assessed using the Bartlett chi-square statistic to determine if the Pearson correlation coefficient between the individual differences (M1 and M2) and the average value [(M1 + M2)/2] was significant (M1 and M2 are tumor volume estimates by two different imaging modalities). 738

3 Ultrasound and brain tumor surgery TABLE 1 Efficacy of intraoperative ultrasound for various assessments during brain tumor surgery in 70 patients No. of Patients & Tumor Status No Previous Radiation- Previous Surgery &/or Induced Metastatic Uses Therapy Radiation Changes Tumors localization well localized poorly localized margin definition well defined moderately defined poorly defined extent of resection well defined poorly defined total patients Results The efficacy of IOUS in localizing tumors, defining tumor borders, and assessing the extent of resection was determined separately for primary gliomas, recurrent gliomas, tumors with radiation-induced changes, and metastatic tumors. The results are listed in Table 1. Localization of Tumors Of the 36 glioma patients, 18 had no previous therapy. The tumors were well localized using IOUS in all 18 (100%) of these patients. The remaining 18 glioma patients had undergone previous surgery and/or radiation therapy; five of these had recurrent tumors and 13 had radiation-induced lesions. All five (100%) recurrent tumors were well localized by IOUS. In the 13 patients for whom pathology showed mainly radiation-induced changes, the lesions were well localized in eight (62%) and poorly localized in five (38%). All 34 metastatic lesions were well localized. Definition of Tumor Borders Of the 18 primary gliomas, definitions of margins were well defined in 15 (83%), and moderately defined in three (17%); in no case was the margin poorly defined. For the five tumors that were recurrent, margins were well and moderately defined in three and two (60% and 40%, respectively). For the 13 with radiation-induced lesions, the margins were moderately defined in two (15%) and poorly defined in 11 (85%). All 34 metastatic lesions had welldefined margins on IOUS. Extent of Resection In all 18 (100%) patients with primary gliomas, the extent of resection (total, 13 patients; subtotal, five patients) was well defined on IOUS, as confirmed by measurements taken on postoperative MR imaging (p = 0.90). For the five patients with recurrent gliomas, the extent of resection was well defined in four (80%) and poorly defined in one (20%). The extent of resection for the former group (total, three; subtotal, one) was confirmed by FIG. 4. Scatterplot showing preoperative tumor volumes on magnetic resonance imaging versus preexcision intraoperative ultrasound volumes for all tumors grouped together. MR imaging (p = 0.59). For the 13 patients with radiationinduced changes, the extent of resection was poorly defined in all (100%) of these patients. The extent of resection was well defined by IOUS images in all cases of metastatic tumor, results that were confirmed with postoperative MR imaging (mean difference = 0). Pre- and Postoperative Tumor Volumes Preexcision tumor volumes measured using IOUS tended to be larger than those defined preoperatively by MR imaging in all cases grouped together (Fig. 4) (mean difference = 1.34 cc; p = 0.025). Mean differences by tumor type were as follows: 1) low-grade gliomas (mean difference = 4.85 cc; p = 0.007); 2) high-grade gliomas (mean difference = 2.19 cc; p = 0.20); and 3) metastatic lesions (mean difference = 0.06 cc; p = 0.92) (Fig. 5). Postexcision tumor volumes assessed using IOUS were significantly correlated with those revealed postoperatively on MR images in all cases grouped together (mean difference = cc; p = 0.97): low-grade gliomas (mean difference = 0.32 cc; p = 0.20); high-grade gliomas (mean difference = 0.21 cc; p = 0.57). In metastatic lesions, the mean difference was zero (Fig. 5). FIG. 5. Graph comparing pre- and postoperative mean tumor volumes revealed by intraoperative ultrasound and magnetic resonance imaging. HGG = high-grade gliomas; LGG = low-grade gliomas; METS = brain metastases. 739

4 M. A. Hammoud, et al. FIG. 6. Graph showing postoperative tumor volumes on magnetic resonance imaging versus postexcision volumes on intraoperative ultrasound in patients with residual tumors. Postoperative MR images revealed that nine patients had residual tumors (volume 0) (Fig. 6). Although postexcision IOUS showed residual tumors in only six of these patients, the difference was not statistically significant (mean difference = 0.02 cc; p = 0.97). Discussion Appropriate neurosurgical intervention often offers the patient with a brain tumor an improved quality of life, prolonged survival, and improved control of neurological problems, even in cases of extremely aggressive tumors. The extent of resection plays a crucial role in the degree to which these advantages can be expected. Optimum resection can be achieved when the tumor is specifically localized, the borders are clearly elucidated, and any residual tumor is readily identified. Intraoperative techniques offer the obvious advantage of determining these factors in real time. 5,9,12,13,23 Uses of IOUS Ultrasound has been shown to be especially efficacious for several functions, namely localizing, defining borders of, and differentiating the tumor from cyst or necrosis; 8,14,28,29 guiding the surgeon; and detecting residual tumor. In this prospective study of 70 patients, we used IOUS to localize the tumors and define their margins and compared pre- and postoperative MR images with IOUS images to evaluate the efficacy of IOUS for assessing the extent of resection. Localization. The principal use of IOUS imaging is for localization. Both low- and high-grade gliomas, as well as metastatic tumors, are readily identified by IOUS imaging and are typically echogenic relative to the surrounding brain 7,9,17,18,30 (Fig. 1). Autopsy studies of gliomas indicate that this echogenicity is related to tumor cell density and extracellular constituents. 24 Our results showed IOUS to be 100% effective in localizing 1) gliomas in patients who had not undergone previous surgery (18 patients); 2) metastatic tumors (34 patients); and 3) recurrent tumors in patients who underwent previous surgery and/or radiation (five patients). For the 13 patients who had radiation-induced changes, IOUS was less precise, a finding that has not received prior recognition. Definition of Borders. The appearance and consistency of high-grade gliomas are usually different from those of normal tissue at the time of surgery, but the actual margins of these tumors cannot be absolutely ascertained. Reportedly, IOUS delineates both gliomas and their transition toward normal tissue, regardless of their CT or MR imaging patterns, 12,16,23 and edema from solid tumor and normal brain, which CT and MR imaging cannot achieve. 4,10 Le Roux, et al., 17 compared IOUS with preoperative CT and MR imaging studies of 33 patients with low-grade gliomas and showed that these tumors are readily identified as hyperechoic and their margins are well defined on IOUS, regardless of the results of preoperative studies. Enzmann, et al., 7 in a study of 41 intracranial tumors, reported that margins of nonenhanced primary astrocytomas were revealed by IOUS but not by CT and that IOUS revealed sharp margins in metastases. In contrast, a study of 76 gliomas of all grades showed that the borders of gliomas could not be discovered by IOUS and that on occasion, the borders of low-grade gliomas were almost undefinable. 33 Auer and van Velthoven 3 also reported that borders were less clear on IOUS than CT, although in 44 of 73 patients investigated, a difference could be seen between the hyperechoicity of the tumor border and that of surrounding edema. Our study is more in agreement with those of Le Roux, et al., and Enzmann, et al. In addition, it highlights the fact that radiation-induced lesions tend to have poorly defined margins. Extent of Resection. To date, the preferred method of determining the extent of resection is postoperative MR imaging. One study has reported that IOUS was unreliable for detecting residual tumor at the end of the operation. 33 In contrast, Rubin and Dohrmann 27 have reported in a study of 191 patients who underwent surgical procedures between 1980 and 1984 that IOUS aided in detecting residual tumor after complete resection was attempted; however, no correlation with postoperative studies was conducted. In our study, IOUS was able to define well the extent of resection in all 18 patients with gliomas and no previous surgery, in four of the five patients with recurrent gliomas, and in all 34 patients with metastatic tumors. In contrast, in all patients who harbored radiation-induced lesions, the extent of resection was poorly defined. To our knowledge, our study is the first to assess the viability of ultrasound in detecting residual tumor by comparing postexcision tumor volumes measured using IOUS with those defined postoperatively by MR imaging. Our data show that tumor volumes measured using postexcision IOUS and MR imaging were significantly correlated for gliomas (p = 0.97) and that the mean difference for metastases was zero. Finally, preexcision tumor volumes of gliomas measured by IOUS tended to be larger than those seen on preoperative MR images (Fig. 5); the largest mean difference occurred with the low-grade gliomas (p = 0.007). These preexcision results are consistent with those of Le Roux, et al., 18 and McGahan, et al

5 Ultrasound and brain tumor surgery Conclusions Our results show that IOUS is extremely useful in localizing tumors and defining margins of gliomas and metastatic lesions, and also demonstrate, by comparing it with MR images, that IOUS is effective in determining the extent of resection of these tumors. Our study is the first to show that the results of postexcision ultrasound and postoperative MR imaging are so closely correlated that, when the extent of resection has been properly determined by ultrasound, postoperative MR imaging for this determination may be unnecessary. If this is the case, this additional capability of IOUS proffers a very cost-effective alternative for determining residual tumor. Acknowledgment The authors thank Ms. Ursula Steinkoenig for her assistance with the preparation of the manuscript. References 1. Altman DG, Bland JM: Measurement in medicine: the analysis of method comparison studies. Statistician 32: , Ammirati M, Vick N, Liao YL: Effect of the extent of surgical resection on survival and quality of life in patients with supratentorial glioblastomas and anaplastic astrocytomas. Neurosurgery 21: , Auer LM, van Velthoven V: Intraoperative ultrasound (US) imaging. Comparison of pathomorphological findings in US and CT. Acta Neurochir 104:84 95, Brant-Zawadzki M, Badami JP, Mills CM, et al: Primary intracranial tumor imaging: a comparison of magnetic resonance and CT. Radiology 150: , Chandler WF, Knake JE: Intraoperative use of ultrasound in neurosurgery, in Weiss MH (ed): Clinical Neurosurgery. Proceedings of the Congress of Neurological Surgeons. Chicago, Illinois Baltimore: Williams & Wilkins, 1983, Vol 31, pp Chandler WF, Knake JE, McGillicuddy JE, et al: Intraoperative use of real-time ultrasonography in neurosurgery. J Neurosurg 57: , Enzmann DR, Wheat R, Marshall WH, et al: Tumors of the central nervous system studied by computed tomography and ultrasound. Radiology 154: , Gooding GAW, Boggan JE, Weinstein PR: Characterization of intracranial neoplasms by CT and intraoperative sonography. AJNR 5: , Gooding GAW, Edwards MSB, Rabkin AE, et al: Intraoperative real-time ultrasound in the localization of intracranial neoplasms. Radiology 146: , Johnson PC, Hunt SJ, Drayer BP: Human cerebral gliomas: correlation of postmortem MR imaging and neuropathologic findings. Radiology 170: , Kelly PJ, Daumas-Duport C, Kispert DB, et al: Imaging-based stereotaxic serial biopsies in untreated intracranial glial neoplasms. J Neurosurg 66: , Knake JE, Chandler WF, Gabrielsen TO, et al: Intraoperative sonographic delineation of low-grade brain neoplasms defined poorly by computed tomography. Radiology 151: , Knake JE, Chandler WF, McGillicuddy JE, et al: Intraoperative sonography for brain tumor localization and ventricular shunt placement. AJR 139: , Koivukangas J, Louhisalmi Y, Alakuijala J, et al: Ultrasoundcontrolled neuronavigator-guided brain surgery. J Neurosurg 79:36 42, Laws ER Jr, Taylor WF, Bergstralh EJ, et al: The neurosurgical management of low-grade astrocytoma. Clin Neurosurg 33: , LeRoux PD, Berger MS, Ojemann GA, et al: Correlation of intraoperative ultrasound tumor volumes and margins with preoperative computerized tomography scans. An intraoperative method to enhance tumor resection. J Neurosurg 71: , Le Roux PD, Berger MS, Wang K, et al: Low grade gliomas: comparison of intraoperative ultrasound characteristics with preoperative imaging studies. J Neurooncol 13: , Le Roux PD, Winter TC, Berger MS, et al: A comparison between preoperative magnetic resonance and intraoperative ultrasound tumor volumes and margins. J Clin Ultrasound 22: 29 36, Leavens ME, Moser RP, Obbens EAMT, et al: Surgical treatment of metastatic brain tumors. Cancer Bull 38:39 44, Lilja A, Bergström K, Spännare B, et al: Reliability of computed tomography in assessing histopathological features of malignant supratentorial gliomas. J Comput Assist Tomogr 5: , Lunsford LD, Martinez AJ, Latchaw RE: Magnetic resonance imaging does not define tumor boundaries. Acta Radiol Suppl 369: , Mahaley MS Jr, Mettlin C, Natarajan N, et al: National survey of patterns of care for brain-tumor patients. J Neurosurg 71: , McGahan JP, Ellis WG, Budenz RW, et al: Brain gliomas: sonographic characterization. Radiology 159: , Medical Research Council Brain Tumour Working Party: Prognostic factors for high-grade malignant glioma: development of a prognostic index. J Neurooncol 9:47 55, Patchell RA, Cirrincione C, Thaler HT, et al: Single brain metastases: surgery plus radiation or radiation alone. Neurology 36: , Quencer RM, Montalvo BM: Intraoperative cranial sonography. Neuroradiology 28: , Rubin JM, Dohrmann GJ: Efficacy of intraoperative US for evaluating intracranial masses. Radiology 157: , Rubin JM, Dohrmann GJ: Intraoperative neurosurgical ultrasound in the localization and characterization of intracranial masses. Radiology 148: , Rubin JM, Mirfakhraee M, Duda EE, et al: Intraoperative ultrasound examination of the brain. Radiology 137: , Sjölander U, Lindgren PG, Hugosson R: Ultrasound sector scanning for the localization and biopsy of intracerebral lesions. J Neurosurg 58:7 10, Sundaresan N, Galicich JH, Beattie EJ Jr: Surgical treatment of brain metastases from lung cancer. J Neurosurg 58: , Tobler WD, Sawaya R, Tew JM Jr: Successful laser-assisted excision of a metastatic midbrain tumor. Neurosurgery 18: , van Velthoven V, Auer LM: Practical application of intraoperative ultrasound imaging. Acta Neurochir 105:5 13, Voorhies RM, Engel I, Gamache FW Jr, et al: Intraoperative localization of subcortical brain tumors: further experience with B-mode real-time sector scanning. Neurosurgery 12: , White KT, Fleming TR, Laws ER Jr: Single metastasis to the brain. Surgical treatment in 122 consecutive patients. Mayo Clin Proc 56: , 1981 Manuscript received September 7, Accepted in final form December 12, This study was supported in part by the Anthony Donald Bullock III Memorial Research Fund. Address reprint requests to: Raymond Sawaya, M.D., Department of Neurosurgery, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe, Box 64, Houston, Texas