Stereotactic radiosurgery of small intracranial tumors: neuropathological correlation in three patients

Effects of effective stereotactic radiosurgery for brain metastases on the adjacent brain parenchyma

Background

To evaluate whether functional and metabolic MRI can detect radiation-induced alterations in the adjacent areas after effective stereotactic radiosurgery (SRS) for brain metastases. If confirmed, these techniques may be suited for monitoring the timely stratification of patients for neuroprotective treatments after irradiation.

Methods

Inclusion criteria were complete response, partial response, or stable disease on routine follow-up MR-scans. Multiparametric 3T-MRI was performed with diffusion-weighted imaging, dynamic susceptibility perfusion-weighted imaging, and two-dimensional proton MR-spectroscopy. Parameters were measured in the SRS-treated target and in the adjacent parenchyma up to both 0.75 cm and 1.5 cm from the target border.

Results

Nineteen lesions in sixteen consecutive patients met the inclusion criteria. The median follow-up time was 39 months (range, 10–142) with 41 multiparametric MR-examinations in total. We found low values of N-acetyl-aspartate up to 1.5 cm from the target borders of SRS (P = 0.043) associated with high values of choline (P = 0.004) at the end of the observation period. Lactate levels in the adjacent tissue declined over time, whereas continuously high apparent-diffusion-coefficient values were noted (P < 0.001).

Conclusion

Multiparametric MRI can depict radiobiological effects and their time course at a distance from the effectively treated site after SRS for brain metastases, even if conventional MRI findings are inconspicuous.

Early and Late Complications Following Dynamic Stereotactic Radiosurgery and Fractionated Stereotactic Radiotherapy

Between December 1986 and June 1990, 112 patients (116 lesions), underwent treatment with dynamic stereotactic radiosurgery at McGill University. Of the treated lesions, 59 were arteriovenous malformations and 53 were a variety of other neoplastic or non-neoplastic conditions. In 86 lesions, the treatment was delivered in a single fraction and the treatment of the remaining 30 lesions was fractionated. Complications attributed to treatment developed in seven of the 112 patients (6.3%). No relationship was found between complications and prescribed dose, fractionation, collimator diameter, type and anatomical region of the lesion that was treated, or previous irradiation. Although extensive clinical experience will be necessary to determine optimal total doses, the potential role of fractionated treatment, and the tolerance of critical structures to radiosurgery, the relatively low incidence of complications in our series allows us to conclude that radiosurgery is well tolerated by the vast majority of patients.

Cerebral metastases pathology after radiosurgery

BACKGROUND

To the authors' knowledge, comprehensive human pathologic investigations that explore fundamental radiosurgical effects on metastatic brain tumors are sparse in the literature. The objective of this study was to analyze histopathologic findings in a set of clinically recurrent cerebral metastases after patients underwent stereotactic radiosurgery (SRS).

METHODS

In a series of 7500 patients who underwent radiosurgery, 2020 patients (27%) harbored cerebral metastases. Eighteen of 2020 patients (0.9%) underwent subsequent craniotomy for tumor removal anywhere from 1 month to 59 months after they received high-dose irradiation. Histologic and immunohistochemical investigations were performed on the surgically resected tissue specimens. These specimens were within the radiosurgical treatment volume of the metastatic tumor.

RESULTS

Light microscopy revealed 3 basic categories of histologic responses: acute-type, subacute-type, and chronic-type tissue reactions. A moderate-to-intense inflammatory cell reaction was seen in the tissue responses of well controlled neoplasms (i.e., in patients who had neoplasms that required craniotomy for recurrent disease > 5 months after SRS), whereas the inflammatory reaction was missing or sparse in poorly controlled neoplasms (patients who required craniotomy for recurrent disease < 5 months after SRS). This reaction was seen within the irradiated tumor volume and not in the peritumoral area nor in areas remote from the radiosurgical treatment volume. Immunohistochemical characterization demonstrated the presence of prominent CD68-positive macrophage and CD3-positive T-lymphocyte populations. A progressively severe vasculopathy also was observed with increasing time after radiosurgery.

CONCLUSIONS

Although causality has not been established, a brisk inflammatory response and more severe vasculopathy were observed in lesions in which recurrences were more delayed.

Evaluation of the response of metastatic brain tumors to stereotactic radiosurgery by proton magnetic resonance spectroscopy, 201TlCl single-photon emission computerized tomography, and gadolinium-enhanced magnetic resonance imaging

Object. The goal of this study was to investigate the usefulness of proton (1H) magnetic resonance (MR) spectroscopy to evaluate the response of metastatic brain tumors to stereotactic radiosurgery (SRS) in comparison with Gd-enhanced MR imaging and single-photon emission computerized tomography with administration of thallium-201 chloride (201TlCl-SPECT).

Methods. Forty patients with a total of 47 metastatic brain tumors were evaluated. The primary lesion was identified in all cases. Stereotactic radiosurgery was effective in 37 lesions. All patients were examined using Gd-enhanced MR imaging before and after SRS. Thalium-201 chloride was administered to 27 patients with 34 tumors and SPECT images were obtained. Proton MR spectroscopy was performed in 36 patients who harbored 43 tumors. On Gd-enhanced MR images, a decrease in the volume of the Gd-enhanced lesion and a change in the enhanced effect in the lesion after treatment were recognized as showing the effectiveness of SRS between 1 and 3 months or more (mean 8.54 ± 3.58 weeks). In 201TlCl-SPECT studies, the ratio of lesion to normal brain decreased from 2 weeks to 2 months (mean 5.03 ± 2.77 weeks) after radiosurgery. On 1H-MR spectroscopy images a high choline (Cho) peak and a lipiddominant (Lip) peak were observed in 25 lesions and a high Cho peak and a lactate-dominant (Lac) peak were observed in 12 lesions before SRS. A decrease in the Cho peak, a disappearance of the Lac peak, and an increase in the Lip peak were observed between 1 week and 1 month (mean 2.76 ± 1.62 weeks) after treatment.

Conclusions. Based on histopathological findings obtained at autopsy or at surgery, we assume that a high Cho peak may be observed in viable tumor tissue and a Lip peak in areas of necrosis. The results indicate that 1H-MR spectroscopy is potentially a more sensitive tool in evaluating the response to SRS than 201TlCl-SPECT or Gd-enhanced MR imaging and that it can be used earlier for this purpose than those other imaging methods.

Cranial nerve preservation after radiosurgery of vestibular schwannomas

Radiosurgery is a management approach used to treat patients with vestibular schwannomas. The goals are long-term tumour growth control, maintenance of cranial nerve function and prevention of new deficiencies. We sought to determine long-term outcomes measuring the potential benefits against the neurological risks of primary radiosurgery. Gamma Knife radiosurgery was applied as a treatment modality for 289 patients with vestibular schwannomas from April 1992 to April 2002. The long-term results of 100 patients who underwent radiosurgery were evaluated. 60 patients received a primary treatment, 40 other cases presented with previously performed subtotal microsurgical resection or recurrence of disease (12-96 months, median 39). The median treatment volume was 3.4 ccm and the median dose to the tumour margin was 13 Gy. The median patient follow-up time was 76 months (range 60-120 months). Four tumours progressed after primary radiosurgery. Tumour control rate was 96%. Useful hearing (Gardner-Robertson I/II) was preserved in 16 patients (55%). Clinical neurological improvement occurred in 50%. Adverse effects comprised neurological symptoms (incomplete facial palsy) (House-Brackman II/III) in six cases (four recovered completely), mild transient trigeminal neuropathy in five cases, and morphological changes displaying rapid enlargement of preexisting macrocysts in two patients and tumour growth in two other patients. Microsurgical resection was performed in four cases (4%) and two patients underwent a shunting procedure because of hydrocephalus formation (2%). In patients who had undergone previous microsurgery, no new cranial nerve deficit was observed. Radiosurgery is an effective method for growth control of vestibular schwannomas and is associated with both a low mortality rate and a good quality of life. Accordingly, for the preservation of cranial nerve function radiosurgery is a useful method for the management of properly selected patients and is comparable to microsurgery.

Effects of stereotactic radiosurgery on metastatic brain tumors of various histopathologies

Although reports have been published describing clinical results in a large series of patients with metastatic brain tumors treated by stereotactic radiosurgery (SRS), clinical neuropathological correlation has rarely been available. The present paper describes three autopsy cases and one surgical case treated with linear accelerator based radiosurgery. The cases comprised a lung cancer, a rectal cancer, an osteosarcoma, and a malignant melanoma. Histological sections of each tumor were analyzed by light microscopy based on the Ohosi and Shimosato's histopathological classification of the effects of radiation therapy. In three cases (pulmonary squamous cell carcinoma, rectal adenocarcinoma and osteosarcoma), a large area of the tumors consisted of coagulation necrosis and non-viable tumor cells, while coagulation necrosis and non-viable tumor cells comprised a very small part of the malignant melanoma. Histopathological type of the metastatic brain tumor may be one of the factors influencing outcome after SRS.

Stereotactic radiosurgery versus stereotactic radiotherapy for patients with vestibular schwannoma: a Leksell Gamma Knife Society 2000 debate

✓ By definition, the term “radiosurgery” refers to the delivery of a therapeutic radiation dose in a single fraction, not simply the use of stereotaxy. Multiple-fraction delivery is better termed “stereotactic radiotherapy.” There are compelling radiobiological principles supporting the biological superiority of single-fraction radiation for achieving an optimal therapeutic response for the slowly proliferating, late-responding, tissue of a schwannoma. It is axiomatic that complication avoidance requires precise three-dimensional conformality between treatment and tumor volumes. This degree of conformality can only be achieved through complex multiisocenter planning. Alternative radiosurgery devices are generally limited to delivering one to four isocenters in a single treatment session. Although they can reproduce dose plans similar in conformality to early gamma knife dose plans by using a similar number of isocenters, they cannot reproduce the conformality of modern gamma knife plans based on magnetic resonance image—targeted localization and five to 30 isocenters.

A disturbing trend is developing in which institutions without nongamma knife radiosurgery (GKS) centers are championing and/or shifting to hypofractionated stereotactic radiotherapy for vestibular schwannomas. This trend appears to be driven by a desire to reduce complication rates to compete with modern GKS results by using complex multiisocenter planning. Aggressive advertising and marketing from some of these centers even paradoxically suggests biological superiority of hypofractionation approaches over single-dose radiosurgery for vestibular schwannomas. At the same time these centers continue to use the term radiosurgery to describe their hypofractionated radiotherapy approach in an apparent effort to benefit from a GKS “halo effect.” It must be reemphasized that as neurosurgeons our primary duty is to achieve permanent tumor control for our patients and not to eliminate complications at the expense of potential late recurrence. The answer to minimizing complications while maintaining maximum tumor control is improved conformality of radiosurgery dose planning and not resorting to homeopathic radiosurgery doses or hypofractionation radiotherapy schemes.

Risk of injury to cranial nerves after gamma knife radiosurgery for skull base meningiomas: experience in 88 patients

OBJECT

In this study the authors sought to determine the neurological risks and potential clinical benefits of gamma knife radiosurgery for skull base meningiomas.

METHODS

A consecutive series of 88 patients harboring skull base meningiomas were treated between 1990 and 1996 by using the Leksell gamma knife in a prospective clinical study that included a strict dose-volume protocol. Forty-nine patients had previously undergone surgery, and six had received external-beam radiotherapy. The median treatment volume was 10 cm3, and the median dose to the tumor margin was 16 Gy. The radiosurgical dosage to the optic nerve, the cavernous sinus, and Meckel's cave was calculated and correlated with clinical outcome. The median patient follow-up time was 35 months (range 12-83 months). Two tumors (2.3%) progressed after radiosurgery; the progression-free 5-year survival rate was 95%. At last follow-up review, 60 (68%) tumors were smaller and 26 (29.5%) remained unchanged. Clinical improvement (in vision, trigeminal pain, or other cranial nerve symptoms) occurred in 15 patients. Functioning optic nerves received a median dose of 10 Gy (range 1-16 Gy), and no treatment-induced visual loss occurred. Among nine patients with new trigeminal neuropathy, six received doses of more than 19 Gy to Meckel's cave.

CONCLUSIONS

Gamma knife radiosurgery appeared to be an effective method to control the growth of most skull base meningiomas in this intermediate-term study. The risk of trigeminal neuropathy seemed to be associated with doses of more than 19 Gy, and the optic apparatus appeared to tolerate doses greater than 10 Gy. Considering the risks to cranial nerves associated with open surgery for comparable tumors, the authors believe that gamma knife radiosurgery is a useful method for the management of properly selected recurrent, residual, or newly diagnosed skull base meningiomas.

Death from a malignant cerebellopontine angle triton tumor despite stereotactic radiosurgery. Case report

Malignant vestibular nerve tumors are rare: to date, only three cases have been reported in the literature. The authors report a case of an eighth cranial nerve tumor that progressed 5 years after stereotactic radiosurgery. The patient was a 44-year-old man who underwent stereotactic radiosurgery for a 27-mm cerebellopontine angle tumor that was discovered on investigation of tinnitus and hearing loss. He developed facial weakness after 5 years, and repeated imaging revealed tumor enlargement. Despite complete microsurgical excision, the tumor rapidly recurred locally and subsequently disseminated within the neuraxis. The patient died 1 year after tumor progression was detected. Histopathological analysis revealed a malignant spindle cell neoplasm with frequent mitotic figures. The presence of positive rhabdoid elements on immunohistochemical studies confirmed that it was a triton tumor. The authors review the relevant literature concerning the classification and management of malignant vestibular nerve tumors and discuss the implications of tumor progression after stereotactic radiosurgery.

Stereotactic radiosurgery for glioblastoma: a final report of 31 patients

From February 1989 to December 1992, 31 patients who presented with an initial pathological diagnosis of glioblastoma multiforme underwent tumor debulking or biopsy, stereotactic radiosurgery, and standard radiation therapy as part of their primary treatment. Presenting characteristics in the 22 men and nine women included a median age of 57 years, Karnofsky Performance Scale score median of 80, and median tumor volume of 16.4 cm3. Stereotactic radiosurgery delivered a central dose of 15 to 35 Gy with the isocenter location, collimator size, and beam paths individualized by means of three-dimensional software developed at the University of Wisconsin. The peripheral isodose line varied from 40% to 90% with a median of 72.5% and a mode of 80%. The mean follow-up period was 12.84 months with a median of 9.5 months. Statistical analysis was performed using Kaplan-Meier analysis and log-rank comparison of risk factor groups. The parameters of age, initial Karnofsky Performance Scale score, and biopsy were significantly different in patient survival from debulking; but no difference was noted between single and multiple isocenters and patterns of steroid requirement. Radiographic recurrences were divided by location into the following categories: central (within central stereotactic radiosurgery dose), 0; peripheral (within 2 cm of central dose), 19; and distant (> 2 cm), 4. There is no evidence of recurrence in five surviving patients. Actuarial 12-month survival was 37%, with a median survival of 9.5 months. These values are similar to previous results for surgery and standard radiotherapy alone. The results suggest that the curative value of radiosurgery is significantly limited by peripheral recurrences.

The radiobiology of human acoustic schwannoma xenografts after stereotactic radiosurgery evaluated in the subrenal capsule of athymic mice

An experimental model with xenograft transplantation into the subrenal capsule of athymic (nude) mice was used to evaluate the early response of human acoustic schwannomas to stereotactic radiosurgery. After xenograft placement, 45 mice underwent radiosurgery with single doses of 10, 20, or 40 Gy using a 201-source 60Co gamma unit (4-mm collimator, single isocenter, 80% isodose line). The 45 radiosurgery-treated xenografts were compared with 15 untreated xenografts and 15 xenografts in mice that underwent "sham radiosurgery." All five study groups were matched for the following pretreatment variables: patient of origin, animal weight, average xenograft diameter, and percentage of xenograft surface vascularity. Immediately prior to sacrifice of the mice all xenografts were evaluated in situ to determine the average tumor diameter, tumor volume, and percentage of surface vascularity. Mice were sacrificed 2 weeks, 1 month, or 3 months after radiosurgery. Blinded histological review was performed by an independent neuropathologist. Tumor volume was reduced 33.6% after 2 weeks (p = 0.023) and 45% after 3 months (p = 0.018) in the 40-Gy radiosurgery group. Tumor volume was reduced by 46.2% after 1 month (p = 0.0002) and 35.2% after 3 months (p = 0.032) in the 20-Gy radiosurgery group. An average volume reduction of 16.4% was observed after 3 months (p = 0.17) in the 10-Gy radiosurgery group. At 3 months after surgery, tumor surface vascularity was reduced by an average of 19.7% (p = 0.043) in the 40-Gy radiosurgery group and 5.8% (p = 0.12) in the 20-Gy radiosurgery group and was unchanged in the 10-Gy radiosurgery group and both control groups. Histological examination demonstrated a higher incidence of hemosiderin deposits (p = 0.026) and vascular mural hyalinization (p = 0.032) in radiosurgery xenografts versus control. The subrenal capsule xenograft in nude mice was an excellent model for studying the in vivo radiobiology of acoustic schwannomas after radiosurgery. Both cellular and vascular effects could be assessed serially in situ and the model was stable even 4 months after transplantation. Additional studies investigating radiobiology over periods better approximating the time course of clinical neuroimaging changes (6 to 12 months) are warranted.

Radiobiology of radiosurgery: Part I. The normal rat brain model

Because limited histological information is available from clinical radiosurgical experience, animal investigations are needed to answer questions regarding the biological response of both normal and pathological tissues. To determine the radiosurgical dose-response relationship of normal brain, we irradiated the right frontal lobe of 18 rats with a single 4-mm isocenter of stereotactic irradiation using the 201-source 60Co gamma unit. Maximal single-fraction doses varied from 30 to 200 Gy (2 rats per dose). All animals were observed for 90 days, killed, and histologically examined. No animal developed neurological dysfunction during that interval, regardless of dose. Animals that received 30, 40, 50, or 60 Gy had no pathological changes. In those given 70 Gy, we found occasional shrunken neurons, and at 80 Gy, rare arteriolar wall thickening. One animal that received 100 Gy had marked capillary endothelial cell degeneration and protein extravasation in the target volume, and the other had a 4-mm diameter necrotic region. Circumscribed cerebral necrosis also was identified in all 4 rats treated with either 150 or 200 Gy; astrocytosis, edema, and microhemorrhage were noted within the surrounding 1 to 2 mm of adjacent brain, and tissue outside that volume had a more normal appearance. We constructed a dose-response relationship based on the cellular, spatial, and temporal effects of focused single-fraction irradiation of the rat brain. To determine the temporal evolution of a known necrotic lesion (200 Gy), 12 other animals were killed (2 each) 1, 7, 14, 21, 30, or 60 days after radiosurgery.(ABSTRACT TRUNCATED AT 250 WORDS)

Radiobiology of radiosurgery: Part II. The rat C6 glioma model

We developed an experimental animal model to evaluate the potential role of stereotactic radiosurgery for glial neoplasms. Rats were randomized to control or treatment groups after implantation of C6 glioma cells into the right frontal region; 14 days later, 19 rats underwent stereotactic radiosurgical treatment of the induced tumor, using the 4-mm collimator of the gamma unit. Both groups were observed for up to 65 days after implantation. Treated animals had a mean survival of 39.2 days; the 22 control animals lived a mean of 29.4 days before death from tumor growth (P = 0.07). Six treated animals (32%), but only one control animal, survived the full observation period (P = 0.07). The mean tumor diameter in the control group was 9.64 mm; in the radiosurgery group, it was 6.47 mm (P = 0.001). Compared with tumors in control animals, treated tumors had a hypocellular appearance (P less than 0.001) and demonstrated cellular edema (P less than 0.005) under light microscopy, indicating a direct cytotoxic response to treatment. No difference was identified in the amount of tumor necrosis, intratumor hemorrhage, or degree of brain invasion between the two groups. Variations in the maximum treatment dose (30, 40, 50, 70, or 100 Gy) did not result in observed differences in tumor response. This in vivo rat malignant glioma model is a valuable tool to evaluate the tumoricidal effects of single-fraction, focused irradiation. Additional studies are warranted to evaluate dose-response relationships, radiation sensitizers, and use of radiosurgery with other adjuvant treatments.

Radiosurgery for solitary brain metastases using the cobalt-60 gamma unit: methods and results in 24 patients

To define the role of stereotactic radiosurgery in the treatment of metastatic brain tumors we treated 24 consecutive patients (20 men, 4 women) with the 201-source 60Co gamma unit between May 1988 and March 1990. The primary tumors included malignant melanoma (n = 10), non-small cell lung carcinoma (n = 6), renal cell carcinoma (n = 3), colorectal carcinoma (n = 1), oropharyngeal carcinoma (n = 1), and adenocarcinoma of unknown origin (n = 3). All tumors were less than or equal to 3.0 cm in greatest diameter. Twenty patients received a planned combination of 30-40 Gy whole brain fractionated irradiation and a radiosurgical "boost" of 16-20 Gy to the tumor margins; one patient refused conventional fractionated irradiation. Three patients with recurrent, persistent, or new non-small cell lung carcinomas had radiosurgical treatment 12-20 months after receiving 30-42.5 Gy whole-brain external beam irradiation. Stereotactic computed tomographic imaging was used for target coordinate determination and imaging-integrated dose planning. All tumors were enclosed by the 50-90% isodose shell using one (n = 22), two (n = 1), or three (n = 1) irradiation isocenters. During this 23-month period (median follow-up of 7 months) no patient died from progression of a radiosurgically-treated brain metastasis. Ten patients died of systemic disease (n = 8) or remote central nervous system metastasis (n = 2) between 1 week and 10 months after radiosurgery. One patient had tumor progression and underwent craniotomy and tumor excision 5 months after radiosurgery. To date, median survival after radiosurgery has been 10 months; 1-year survival was 33.3%. Stereotactic radiosurgery eliminated the surgical and anesthetic risks associated with craniotomy and resection of solitary brain metastases. Radiosurgery also effectively controlled the growth of tumors considered "resistant" to conventional irradiation.

Stereotactic radiosurgery of meningiomas

Stereotactic radiosurgery has an expanding role in the management of selected intracranial tumors. In an initial 30-month experience using the 201-source cobalt-60 gamma knife at the University of Pittsburgh, 50 patients with meningiomas were treated. The most frequent site of origin was the skull base. Previously, 36 patients (72%) had undergone at least one craniotomy and four patients (8%) had received fractionated external beam radiation therapy. Stereotactic radiosurgery was the primary treatment modality in 16 patients (32%) with symptomatic tumors demonstrated by neuroimaging. Computer imaging-generated isodose plans (with one to five irradiation isocenters) for single-treatment irradiation gave optimal (greater than or equal to 50% isodose line) coverage in 44 patients (88%). The proximity of cranial nerves or vascular, pituitary, and brain-stem structures to the often convoluted tumor mass was crucial to dose selection. Serial imaging studies were evaluated in all 50 patients. Twenty-four patients were examined between 12 and 36 months after treatment; 13 (54%) showed a reduction in tumor volume while nine (38%) showed no change. Of 26 patients evaluated between 6 and 12 months after treatment, four showed a decrease in tumor size while 22 showed no change. Two patients (both with large tumors that received suboptimal irradiation) had delayed tumor growth outside the radiosurgical treatment volume. The actuarial 2-year tumor growth control rate was 96%. Between 3 and 12 months after radiosurgery, three patients developed delayed neurological deficits that gradually improved, compatible with delayed radiation injury. Although extended follow-up monitoring over many years will be necessary to fully evaluate treatment, to date stereotactic radiosurgery has proved to be a relatively safe and effective therapy for selected patients with symptomatic meningiomas, including those who failed surgical resection. Radiosurgery was an effective primary treatment alternative for those patients whose advanced age, medical condition, or high-risk tumor location mitigated against surgical resection.

Radiosurgery of acoustic neurinomas

Eighty-five patients with acoustic neurinomas underwent stereotactic radiosurgery with the gamma unit at the University of Pittsburgh (Pittsburgh, PA) during its first 30 months of operation. Neuroimaging studies performed in 40 patients with more than 1 year follow-up showed that tumors were smaller in 22 (55%), unchanged in 17 (43%), and larger in one (2%). The 2-year actuarial rates for preservation of useful hearing and any hearing were 46% and 62%, respectively. Previously undetected neuropathies of the trigeminal (n = 12) and facial nerves (n = 14) occurred 1 week to 1 year after radiosurgery (median, 7 and 6 months, respectively), and improved at median intervals of 13 and 8 months, respectively, after onset. Hearing loss was significantly associated with increasing average tumor diameter (P = 0.04). No deterioration of any cranial nerve function has yet developed in seven patients with average tumor diameters less than 10 mm. Radiosurgery is an important treatment alternative for selected acoustic neurinoma patients.

Solitary brain metastasis: radiosurgery in lieu of microsurgery in 32 patients

Thirty-two consecutive patients with 34 small brain metastases underwent boost stereotactic radiosurgery using the first North American Gamma Unit between May 1988 and July 1990. The majority of tumors (n = 24; 71%) were considered resistant to conventional, fractionated irradiation (malignant melanoma, n = 13; non-small cell lung carcinoma, n = 7; renal cell carcinoma, n = 4). During the follow-up period (median = 10 months; range = 1.5-15 months) no patient suffered a complication of radiosurgical treatment, and no patient died from a radiosurgically-treated metastasis. Shrinkage or growth-arrest was documented in 20 of 23 patients (87%) available for follow-up. Median survival after treatment was 10 months.