Stereotactic brain biopsy is considered by many physicians to have significant morbidity and mortality rates with a high risk of sampling error resulting in misdiagnosis. The technical aspects necessary to perform the procedure safely and effectively are unfamiliar to most physicians.
After reporting his initial results with stereotactic brain biopsy, several modifications were implemented by the author to improve the morbidity, mortality, and diagnostic yield rates, including complex surgical planning with regard to patient selection, biopsy trajectory, imaging technique, target choice, and intraoperative pathologic review. The results of implementing these modifications were examined retrospectively in 134 consecutive brain biopsies.
One hundred and thirty-four stereotactic brain biopsies were performed in 122 patients. Computed tomography guidance was used in 85 patients (63%) and magnetic resonance imaging was used in 49 patients (37%). Sixty-four lesions (48%) were located in the right hemisphere, 61 (45%) in the left, and 9 (7%) in the midline. The most common diagnoses included 62 malignant brain tumors (46%), 24 benign brain tumors (18%), 23 neurologic disorders (17%), and 20 infections (15%). Five biopsies (4%) did not demonstrate a pathologic process for an overall diagnostic yield of 96%. Reasons for diagnostic failure included lesion location adjacent to the ventricular system, inaccurate targeting, and the inability to penetrate the tumor. One patient sustained a neurologic deficit after the biopsy for a morbidity rate of 0.7% and one sustained a fatal hemorrhage during the biopsy of a vascular tumor for a mortality rate of 0.7%. These results are comparable to those reported in 7471 biopsies (current series included) in which the morbidity rate was 3.5%, the mortality rate was 0.7%, and the diagnostic yield was 91%.
Stereotactic biopsy allows neurosurgeons to localize and sample intrinsic lesions located anywhere within the brain accurately. With the advent of computed tomography (CT) in the 1970s came the ability to precisely visualize the location of lesions affecting the central nervous system (CNS). Initially, CT-guided freehand techniques were used to obtain tissue from intracranial lesions until rigidly fixed stereotactic headframes were developed in the early 1980s.1 This procedure is indicated when the pathology of the target is unknown or when future therapy will be influenced by the histologic nature of the lesion. Another indication for the use of this form of surgery is the medically fragile patient in whom a general anesthetic carries substantial risk. Stereotactic sampling is appropriate for lesions that are located deep within the brain, in eloquent cortex, are surgically unresectable and diffusely infiltrating, or are cystic and causing significant mass effect on surrounding neural structures, in which case aspiration may result in decompression of CNS tissue and restoration of function. Stereotactic technique also is useful for determining the etiology of multiple intracranial lesions or when cytoreductive surgery would not benefit the patient.2
In 1993, we reported our initial experience comparing freehand CT-guided needle biopsies with first-generation stereotactic brain biopsies performed over a 6-year period by several different surgeons. For the stereotactic biopsies, the morbidity rate was 6% and the mortality rate was 1.5%. Twelve stereotactic biopsies (18%) were nondiagnostic, which was believed to be due to the small size of the lesions biopsied and the limited amount of tissue obtained at surgery. Because of the considerable morbidity, mortality, and nondiagnostic biopsy rates observed in our early series compared with published results, several modifications were applied to the procedure in an attempt to improve the safety and efficacy. At the time of the biopsy, the tissue was examined pathologically either by frozen section, touch preparation, or smear preparation to confirm the presence or absence of diagnostic tissue.3 Complex preoperative surgical planning was undertaken before performing the procedure in all cases and the results of instituting these modifications were evaluated retrospectively in 134 consecutive stereotactic biopsies.
MATERIALS AND METHODS
The results of 134 consecutive stereotactic biopsies performed in 122 patients between February 1991 and December 1996 at the University of Minnesota Hospital and Clinic were reviewed. Some patients with brain tumors received multiple biopsies over time to evaluate their response to treatment when a change in their therapeutic management was being considered. On occasion, patients with multiple lesions had more than one area sampled if there was concern that more than one pathologic process was present and in some patients in whom the initial biopsy was nondiagnostic the procedure was repeated at a later date. The medical and radiologic records of all patients were analyzed to determine patient age and gender, the location of the lesion, the imaging study used during the biopsy, the frozen-section diagnosis, the permanent pathologic diagnosis, morbidity, mortality, the correlation between the intraoperative and subsequent final diagnoses, and the diagnostic yield of the procedure.
Most biopsies were performed using local anesthesia (1% lidocaine) and intravenous sedation, except in cases in which the lesion was located in the posterior fossa, the temporal lobe, or near the cortical surface where there was a significant risk of vascular injury. Some medial or posterior temporal lesions were approached through the occipital lobe to avoid the blood vessels on the surface of the temporal lobe emanating from the sylvian fissure. Because the procedure is "blind" and the development of an epidural cerebellar or temporal lobe hematoma could result in brain stem compression, lesions in these locations often were biopsied through a burr hole. The burr hole was placed under general anesthesia with direct visualization of the brain surface at the time of passage of the biopsy needle. Burr holes were used for superficial cortical biopsies when it was believed that the development of either an epidural or subdural hematoma would cause significant neurologic injury (i.e., parietal lesion).
The Cosman-Roberts-Wells stereotactic system (Radionics, Burlington, MA) was used for all stereotactic biopsies. After placement of the stereotactic headframe with local anesthesia, a CT or magnetic resonance imaging (MRI) scan of the head was obtained to determine the target coordinates. Placement of the stereotactic head frame was performed under general anesthesia for children and adolescents prior to performing the biopsy. For most patients, an entry point anterior to the coronal suture in the frontal lobe where ventriculostomy catheters usually are placed was chosen for the passage of the biopsy needle because of the safety associated with surgery in this area. For patients with brain stem lesions, MRI guidance was used. These patients had their imaging performed in the coronal plane perpendicular to the head ring. The trajectory of the biopsy needle through the parietal lobe was chosen to avoid the ventricular system and to determine how far laterally to place the burr hole (Fig. 1 (18K)). In the operating room, the 17-gauge Nashold side-aspirating biopsy needle (Radionics) was used to obtain tissue specimens through a twist drill craniostomy or burr hole. We prefer the Nashold biopsy needle over biting forceps or the corkscrew biopsy needle because of a perceived lower risk of vascular injury that can result in hemorrhage and the comparatively larger size of the tissue samples that are obtained. During the procedure, a portion of the sample was selected for pathologic examination to verify the presence of abnormal tissue. In a few cases, a pathologic assessment was not performed because of the small size of the sample. Patients with acquired immunodeficiency syndrome received either a touch or smear preparation of their tissue specimen rather than a formal frozen section to avoid the necessity of cleaning the cryostat immediately after the biopsy. A postoperative CT scan to exclude the presence of hemorrhage rarely was performed. After the biopsy, patients were observed for neurologic deterioration prior to discharge the following morning. Suggested principles for performing a successful stereotactic brain biopsy are listed in Table 1.
Table 1. Principles for Successful Stereotactic Brain Biopsy
Headframe placement in children and adolescents
Burr hole placement
Burr hole placement
Allows direct visual inspection of the cortex
Superficial temporal and parietal lesions
Twist drill craniostomy
Transfrontal approach (preferred)
Transcortical approach for superficial lesions
Parietal approach for pontine lesions
Occipital approach for medial and posterior temporal lesions
Side-aspirating biopsy needle (preferred)
Corkscrew biopsy needle
Patient and Lesion Demographics
The age range for the 122 patients was 3-83 years with a median age of 38 years and a mean age of 41 years. There were 75 men (61%) and 47 women (39%) biopsied. Sixty-four lesions (48%) were on the right side of the brain, 61 (45%) were on the left, and 9 (7%) were in the midline (Table 2). Specific lesion locations are indicated in Table 2. Eighty-five lesions (63%) were biopsied with CT guidance and 49 (37%) with MRI guidance. The majority of lesions in the right hemisphere of the brain were biopsied with CT guidance (n = 47) rather than with MRI guidance (n = 17). Left hemispheric lesions were biopsied slightly more often using CT (n = 34) compared with MRI (n = 27). Midline lesions were biopsied to an almost equal degree with CT (n = 4) or MRI (n = 5) guidance. Intrinsic lesions of the brain stem involving either the midbrain (n = 2) or the pons (n = 2) only were biopsied using MRI. General anesthesia was necessary for the biopsy of 28 lesions (21%). Seven patients (5%) required general anesthesia because of their young age. Patients with lesions located in the pons (n = 2), the cerebellum (n = 5), and adjacent to the cortex, primarily in the temporal and parietal areas (n = 14), also necessitated a general anesthetic.
Table 2. Location of 134 Intracranial Lesions and Biopsy Technique
CT: computed tomography; MRI: magnetic resonance imaging.
The pathologic yield for the lesions biopsied stereotactically is indicated in Table 3.
Table 3. Pathologic Diagnoses of 134 Stereotactically Biopsied Lesions
The term diagnostic yield is used rather than diagnostic accuracy because of the known heterogeneity that exists within brain tumors and the small size of the tissue samples that are obtained with this technique.4, 5 Malignant brain tumors represented the most frequent lesion biopsied (n = 62; 46%), followed by benign tumors (n = 24; 18%), and neurologic disorders (n = 23; 17%). Intraoperative pathologic examination was not performed in 10 patients (7%) to submit all tissue for permanent pathologic sections. Despite intraoperative pathologic evaluation there were 7 samples (5%) that did not demonstrate diagnostic tissue, 4 of which (3%) also were found to be nondiagnostic on permanent pathology.
One hundred and nine patients (89%) had a single biopsy, 10 (8%) had 2 biopsies, and 3 (2%) had 3 biopsies. Of the 10 patients who required two biopsies, six patients had lesion progression over time and there was concern that the pathologic process present represented radiation necrosis rather than recurrent tumor. In all 12 biopsy samples from those 6 patients, recurrent tumor was identified that influenced future therapeutic intervention. There were three patients who each had two distinct lesions that were found to be encephalitis, bacterial abscess, and fungal abscess in both of the biopsied lesions. These three patients had both of their intracranial lesions biopsied because they were bone marrow transplant recipients and they were not responding to either antibacterial and antifungal therapy. One patient had a midbrain lesion, presumed to be recurrent tumor; a biopsy was attempted on two different occasions and was nondiagnostic in both cases because a tissue sample could not be obtained due to the proximity of the lesion to the ventricular system. The two patients undergoing three biopsies (one with an oligodendroglioma, the other with an anaplastic astrocytoma) were followed for several years and biopsied when radiographic or clinical worsening was detected and a change in therapy was contemplated.
There were 10 patients (7%) in whom an intraoperative pathologic examination was not performed either because of the small sample size, the suspected difficulty of diagnosing their neurologic disorder and the need to submit as much tissue as possible for permanent section, and the presence of human immunodeficiency virus (HIV) positivity in the patient. The pathologic diagnoses found in these patients on permanent section were microglial proliferation in three patients and one case each of astrocytoma, anaplastic astrocytoma, inflammation, demyelination, and nonradiation related necrosis. In the two HIV positive patients in this group, there was one case of lymphoma and one Toxoplasma gondii abscess.
There were 7 patients (5%) in whom the tissue examined on frozen section did not demonstrate a pathologic diagnosis and one patient with a pineal mass from whom no tissue was obtained at surgery. Of these seven patients in whom the diagnosis was not apparent on frozen section, a diagnosis was determined in three patients based on the permanent pathologic sections that included an astrocytoma, reactive astrocytosis, and inflammation. Normal brain tissue was observed on both frozen and permanent pathologic examination in the other four patients.
Safety and Efficacy
After the biopsy, 1 patient (0.7%) found to have a pontine anaplastic astrocytoma experienced a temporary hemiparesis due to edema and not hemorrhage that gradually resolved. One patient (0.7%) sustained a fatal hemorrhage at the time of biopsy of a friable, vascular pineal region mass. An emergent craniotomy was performed to evacuate that hematoma at which time it was very difficult to control the intraoperative hemorrhage. In retrospect, the small fragment of tissue obtained in the first pass of the needle was diagnostic for pineoblastoma.
In 7 patients (5%), the initial intraoperative pathologic evaluation did not demonstrate diagnostic tissue, prompting the surgeon to obtain additional tissue for permanent examination, which resulted in increasing the diagnostic yield by 2% (n = 3). Five biopsies (4%) were nondiagnostic. Reasons for diagnostic failure included the proximity of the lesion to the cerebrospinal fluid pathways in two cases in which the samples obtained represented normal tissue adjacent to the ventricle. In two other patients diagnostic tissue was not obtained despite multiple passes at different depths and the procedure was discontinued to avoid a complication even though the intraoperative pathologic review was unremarkable. No tissue was obtained in another patient with a firm pineal region mass because of an inability to penetrate the mass at surgery. The procedure was aborted and an open procedure was performed at which time a teratoma was diagnosed. The overall diagnostic yield for the procedure was 96%.
Stereotactic brain biopsy has been used to sample areas of the brain since the early 1980s. Most neurosurgeons performing this procedure have specialized training in stereotaxis. In reviewing 17 of the largest series of stereotactic brain biopsies that included both our early results and those reported in the current study (Table 4), nearly 7500 procedures have been performed.1, 2, 6-20 The overall mortality rate was 0.7%, the morbidity rate was 3.5%, and the diagnostic yield was 91% for stereotactic biopsies. Because of the discrepancy between our early results and those published in the literature, we instituted several modifications in an attempt to improve diagnostic yield and decrease morbidity and mortality rates. Changes included complex preoperative surgical planning and some form of pathologic assessment was conducted during the procedure in 93% of cases to assure the presence of diagnostic tissue. Repeat sampling was undertaken at the time of surgery at a different depth if the first sample was believed to be nonrepresentative. By initiating these changes, our mortality rate has decreased from 1.3% to 0.7%, the morbidity rate has decreased from 5% to 0.7%, and the diagnostic yield has improved from 80% to 96%. These current mortality, morbidity, and diagnostic yield rates are very comparable to those results reported in the literature.
Factors that have been associated with complications from stereotactic biopsy have been surgical experience and the biology of the lesions undergoing biopsy.2 One of the major reasons cited to explain the wide range of published complication rates has been variability in surgical judgment, experience, and skill.2 Important considerations for the surgeon performing the biopsy include selection of the patient, the target, the trajectory, and the biopsy device.2, 21 Regarding the location of the lesions that were biopsied, the choice of trajectory was believed to be a major factor in complication avoidance with Figure 1 (18K) illustrating the complexity of the treatment planning necessary for some cases. CT guidance was used more often than MRI to biopsy lesions in the right hemisphere because of the absence of language function on that side in the majority of patients. Knowledge, experience, and intellectual interest in stereotactic neurosurgery and/or neurooncology can influence the success of the procedure significantly.
The presence of neovascularization and abnormal blood vessel structure in malignant tumors such as gliomas and lymphomas may explain their propensity to hemorrhage at the time of biopsy and/or produce increased cerebral edema resulting in neurologic deficit after surgical manipulation.2 Reported complication rates for patients with glioma, lymphoma, and metastasis were 6.4%, 6,3%, and 2.8%, respectively.2 The single patient in this series that experienced neurologic worsening had an anaplastic astrocytoma in the brain stem and the sole fatality had a vascular pineoblastoma. In the 40 malignant gliomas identified by permanent pathologic evaluation, the risk of postoperative neurologic deficit was 2.5%. The rate of neurologic deterioration in HIV positive patients ranges from 4-12.5%.2, 22, 23 None of our patients with lymphomas or those who were HIV positive worsened neurologically after the biopsy.
Reasons that have been suggested for diagnostic failure in stereotactic brain biopsy include small sample size, inaccurate tissue targeting resulting in sampling error, target choice in areas of high signal on T2-weighted MRI, and small target size.1 Individual reasons for failure in this group of patients included the inability to penetrate a "hard" pineal region tumor, lesion location adjacent to the ventricles (two cases), and inaccurate tissue sampling despite multiple specimens at several different depths (two cases). Others have recommended that hard tumors in the pineal region probably represent teratomas and should be biopsied under direct vision by craniotomy to avoid complications.6 Intraoperative pathologic evaluation may assure a definitive diagnosis on permanent paraffin sections.1, 3 Applying intraoperative review has increased our diagnostic yield from 80% to 96%, which is comparable to the overall success rate of 91% (range, 80-99%) reported for 7471 stereotactic biopsies. The increased experience with stereotactic procedures and the increasing use of MRI for localization and planning represent possible factors contributing to greater safety and diagnostic yield of the procedure.
Stereotactic brain biopsy is an extremely safe and effective procedure for determining the pathologic nature of intracranial lesions. After careful examination of the reasons for increased morbidity, mortality, and diagnostic failure rates, several modifications were implemented that included complex preoperative surgical planning and intraoperative pathologic review. By instituting these changes, the morbidity and mortality rates each were decreased to 0.7% and the diagnostic yield was increased to 96%. These results and those reported in the literature support the utility of this procedure as an essential tool for neurosurgeons with an interest in stereotactic surgery and neurooncology.