B bidity in cancer patients. Recent data suggest that

Transcription

1 Surgical Treatment of Brain Metastases Clinical and Computerized Tomography Evaluation of the Results of Treatment NARAYAN SUNDARESAN, MD, AND JOSEPH H. GALICICH, MD The results of treatment of brain metastases in a series of patients who underwent surgery with or without postoperative radiation from 978 through 98 were analyzed. The major sites of primary tumor included the lung (%), melanoma-skin (I I%), kidney ( l%), colon (8%), soft tissue sarcoma (8%), breast (6%), and a variety of others (%). At the time of craniotomy, disease was considered limited to the central nervous system in 6 patients (%). After surgery, 8 patients (66%) were neurologically improved, and 6 (%) had their deficits stabilized. The overall median survival was months, and % lived years. Eight patients (%) are alive years or more following surgery. Survival varied with site of primary tumor, location of brain metastasis, extent of systemic disease, and neurologic deficit at time of craniotomy. Over a follow-up period ranging from 8 months to 6 years, patients (%) developed either local recurrences or other sites of brain metastases. These data suggest that although craniotomy followed by radiation is highly effective in the initial treatment of selected patients with brain metastases, alternate therapies require investigation in view of the high central nervous system relapse rate in long-term survivors. Cancer 8-88, 98. RAIN METASTASES represent a major cause of mor- B bidity in cancer patients. Recent data suggest that % to % of patients with cancer have intracranial metastases at autopsy, the majority of which produce symptoms during life.. Some clinicians have noted an increasing frequency of brain metastases complicating the course of the illness in cancer patients, and have speculated that this may be the result of increasing longevity brought about by aggressive t~-eatment.~,~ The current treatment of brain metastases is considered palliative. Approximately % of patients with brain metastases have multiple tumors demonstrated by computerized tomography (CT) scan. Even in patients with single lesions, brain metastases frequently occur in the setting of widespread and uncontrolled systemic cancer. These considerations preclude the possibility of curative therapy in most patients- Therefore, neurologic palliation still remains the most important goal of therapy, even if survival is not substantially altered. For many others who relapse with a single brain metastasis and whose primary tumor appears stable or under control Presented at the 7th Annual Meeting of the Society of Surgical Oncology, New York, New York. May -7, 98. From the Neurosurgery Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, New York. Address for reprints: Narayan Sundaresan MD, 7 York Avenue, New York, NY I. Accepted for publication September 9, 98. by systemic therapy, the best form of therapy continues to be debated. The occasional long-term cures reported in the literature after craniotomy are often used to justify the philosophy of resecting single brain metastasis in selected patients. This approach is conceptually attractive, and is clearly indicated in patients with radioresistant tumors, in others who are symptomatic from large intracranial masses, and in patients who deteriorate acutely from intracranial hemorrhage secondary to a metastatic tumor. Although the reported results of combined-modality therapy (surgery and radiation) are superior to those following treatment by radiation this strategy has yet to be proven superior to radiation therapy alone by prospective randomized studies. More recent reports that compare treatment of patients with single brain metastases from specific sites all seem to suggest an overall survival advantage for those undergoing craniotomy.i6- Unfortunately, only Patchell and have attempted to match patients with similar prognostic variables, whereas in all other studies patients selected for surgery generally had more limited disease and a better performance status. Even the most effective dose-fractionation for those treated by radiation alone is not established, and remains the focus of collaborative ongoing ~tudies.~ -~~ Whether the same dose of whole brain radiation used for palliation should be given postoperatively in patients who have undergone surgical resection of solitary metastasis has recently been q~estioned. ~ More importantly, late con- 8

2 ~ ~ ~~ No. 6 SURGERY FOR BRAIN METASTASES - Sundaresan and Galicich 8 sequences of therapy are becoming apparent, since the cumulative probability of delayed radiation-related complications increases as survival is prolonged. To define the effectiveness of combined modality treatment, we analyzed the results of treatment in patients with brain metastases who underwent surgery at this institution over a -year period. Although this is not a formal prospective study, accurate follow-up data as well as serial CT scans were obtained at regular intervals from the beginning of the study to determine the overall effectiveness of therapy directed to the central nervous The major objective of the study, aside from determining the quality and duration of survival, was to determine central nervous system relapse rates for each solid tumor. We believe that these results may offer guidelines for more appropriate future treatment in patients with single brain metastases. Clinical Material and Methods Over a -year period (978-98, patients underwent surgical resection of their brain metastases at this institution. We have excluded patients with intracranial metastasis from lymphomas and other hematogenous malignancies, as well as those in whom intracranial invasion resulted from direct extension of tumor through the skull or paranasal sinuses. The diagnosis of intracranial metastasis was established by CT scans in all patients. One hundred nineteen patients had single brain metastasis. The remaining six patients initially had two metastases each; following whole brain radiation, only one large symptomatic mass remained, and these patients are therefore also included in this study. The important clinical parameters in this group of patients are shown in Tables and. The median age for the entire group of patients was years; this study includes eight patients in the pediatric age group (younger than years). The brain metastasis was located in the supratentorial compartment in 8% of the patients, and was equally distributed between the right and left hemispheres; in the remaining %, the tumor was located in the posterior fossa. no. of patients Sex: M/F Age range (yr) TABLE I. Clinical Parameters Site of tumor Left hemisphere Right hemisphere Posterior fossa Extent of disease Limited to CNS Extra-CNS disease present Solitary CNS relapse Onset of brain metastases Up to year After months Neurologic grade Grade I Grade I Grade CNS: central nervous system. 69/6.-7 (median, ) (%) (%) 9 (%) 6 (%) 6 (%) 6 (7%) 6 (%) 6 (%) 7 (%) 6 (%) (%) In 6 patients (%), the brain metastasis occurred early in the course of the illness (up to year after diagnosis of the primary cancer). As shown, the majority of patients with early-onset brain metastasis were those with lung cancer. Seventy percent of these patients developed their brain metastasis within year of diagnosis of the primary tumor, compared to % of those with melanoma. In all patients, extent of systemic involvement was evaluated by chest x-ray, as well as radionuclide bone and liver-spleen scans. In the more recent patients, additional evaluation of suspicious lesions noted on radionuclide studies was camed out by CT scans or by ultrasound examinations. Sixty-three patients (%) were believed to have disease limited to the central nervous system. Of these, the brain metastasis represented the first and only site of relapse ( solitary relapse) following apparent curative resection of the primary cancer in 6 patients (7%). Neurologic status at time of craniotomy was classified into three grades, with these assessments being made after full-dose corticosteroid therapy. Grade I patients were those with minimal or no neurologic deficit and who had Karnofsky perfor- TABLE. Clinical Parameters by Primary Site of Tumor Distribution Neurologic grade Primary site (percent) Supra- Infra- Early onset CNS only I I I (%) Skin (melanoma) (.%) I ( I.%) I (8.%) 7 Soft-part sarcoma (8.8) 7 8 (6.%) 7 6 primary site 6 (.8%) Miscellaneous I (.%) I I Supra-: supratentorial: Infra-: infratentorial; CNS: central nervous system; CNS only: disease limited to CNS I 7 8

3 8 CANCER March 98 Vol. Melanoma Sarcoma Miscellaneous TABLE. Radiation Therapy Postoperative Failed WBRT WBRT Others II WBRT: whole brain radiation. I I 8.- >.- VJ.ooo F; c..-.7 a patients, censored median = months % = years mance scales of 8 and more. Grade I patients were those with moderate neurologic deficit and performance scales from to 7. Grade patients were those with severe neurologic deficit and performance scales of less than. Surgery was usually performed after patients had received to days of high-dose corticosteroid therapy. Before surgery, all patients underwent double-dose contrast scans (twice the usual dose of intravenous contrast) to more accurately identify patients with truly single metastasis. In addition, localizing scans were performed in the axial and coronal planes to determine safe surgical approaches to the tumor. The goal of surgery was complete excision of the tumor, which was accomplished in 9% of patients. Completeness of tumor resection was evaluated by postsurgical CT scans done approximately week following craniotomy. Whole brain radiation (WBRT) was then given according to a variety of treatment protocols in effect during this period (Tables and ). Seventy-six patients received WBRT following resection of the brain tumor. In patients, WBRT had been given at varying intervals before craniotomy, and these patients were operated on because they were symptomatic from progressive mass lesions. Of the remaining 8 patients, patients received focal radiation to the tumor bed. Thirteen patients did not receive radiation therapy after surgery. Survival data were calculated in months using the Kaplan-Meier product limit method. We evaluated the following factors for prognostic significance: site of pri- TABLE. Radiation Treatment Protocols Estimated doseffractionfdays TDF rad equivalent dose 9 radf If I7 days 8. radflofl days 6 8 rad//-6 days 67 rad/6/-6 days 89 rad/l/ I days 9 TDF Time-dose fractionation. FIG. I. Overall survival after treatment of brain metastasis. mary tumor, location of brain metastasis, time to onset of brain metastasis, extent of systemic disease, and neurologic grade. In addition, we compared survival of those who received radiation therapy postoperatively with those who had failed previous radiation. The logrank and Breslow tests were used to test for significant differences in survival between subgroups. All analyses were performed with a prepackaged statistical software Neurologic improvement following completion of therapy was assessed, and serial CT scans were obtained on these patients, generally at - to 6-month intervals. Clinical follow-up was possible in all cases, except for three patients who left the country. Results of Treatment Survival data following treatment of the brain metastases are shown in Figures through. The overall median survival was months; % of the patients survived years, and 8 patients (%) are alive years or more after treatment. Survival was dependant on extent of systemic disease at time of craniotomy as well as primary site of original cancer. Median survival for patients with disease limited to the central nervous system was months, and was months in patients with evidence of extracranial disease; this difference was statistically significant (P <. ). Of individual primary tumors, lung cancer patients had the longest median survival; in the remainder, median survival times varied from to months (Table ). Tumor site also affected survival; patients with posterior fossa metastasis had a median survival of 7 months, compared to a median survival of months in patients with supratentorial tumors. This difference was statistically significant with the log-rank test (P <.), but failed to reach significance with the Breslow test. In contrast to our previous data, the survival of patients with early-onset brain

4 No. 6. F; SURGERY FOR BRAIN METASTASES * Sundaresan and Galicich 8 -Late onset : (6 pts, censored) median = months median= months.- P.7.- > ---Early onset: (6 pts, censored) median= 7 months.6 median=lo months.- P >.7b\ ---Posterior fossa: ( 9 pts, censored) $.6 L7 c.- n e a,87 k!l - Supratentorial : (6 pts, censored). O w FIG.. Survival by location of tumor. metastasis was not significantly different from those with late-onset metastasis. However, if patients with lung cancer were excluded, there was a trend toward poorer survival with patients with early-onset metastasis from kidney and other sites. Neurologic grade at time of craniotomy was also associated with survival. Patients with minimal or no neurologic deficit had significantly longer survival times compared with those having moderate or severe neurologic deficit (P <.). Patients with severe neurologic deficits and poor performance status had median survival times of months. There was a close correlation between patients with severe deficit and the presence of extracranial disease in this study. No difference in survival was noted in patients who received postoperative radiation compared with those who had failed previous radiation. The neurologic improvement after surgery is shown in Table. In the overall study, 8 patients (66%) were.ooo - CNS only: (6 pts, censored) median= months, Extra CNS disease present: (6 pts, I censored).- P.7.- > median= 6 months.6 c,.- E.7 a",.i Y-. I I L.+..-[.AI I I I J FIG.. Survival by extent of systemic disease. l.ooor.87 c..- e.7 n g. c I I FIG.. Survival by onset of brain metastasis. Early onset includes diagnosis of brain metastasis made up to I year after diagnosis of cancer. improved, and an additional 6 patients ( %) had their neurologic deficits stabilized. Two thirds of those who improved maintained improvement over I year. Eight patients (6%) died within a -day period after craniotomy, and % of these deaths were due to progression of systemic cancer. Eight patients (6%) experienced increased neurologic deficit postsurgery. An additional eight patients developed postoperative clots that required evacuation. Reexploration was required in four others because the tumor was not successfully localized during the first craniotomy. Fifteen patients ( %) developed other nonfatal complications that delayed discharge from hospital, including urinary tract infections, deep venous thrombosis, or postoperative fever. Central nervous system relapses were classified into three groups: local recurrences at the original site, other.ooo,87.- E'.7.- > $.6 c..- E.7 g..i L- -L-- Normal or minimal deficit: (7 pts, censored) median= months - Moderate neurological deficit: (6 pts, censored) median = months Severe neurological deficit: ( pts, censored) median= 6 months 'L.-I.IL. -. L. -. UL I FIG.. Survival by neurologic grade at time of craniotomy.

5 86 CANCER March 98 Vol. TABLE. Results of Treatment by Primary Site of Tumor Neurologic status Median survival Percent surviving Primary site (mo) Yr Improved Stable Worse Died Skin (Melanoma) Soft part sarcoma Others I sites of intracranial metastases, and patients with both local and distant recurrences. These are shown by tumor type in Table 6. In Table 7, we have tabulated central nervous system relapse rates for those with solitary brain metastases because these patients have the best potential for long-term survival. All central nervous system relapses occurred in the first years following craniotomy. Fifty percent of the -year survivors developed hydrocephalus ex vacuo or evidence of leukoencephalopathy on CT scans. Eight patients showed areas of contrast enhancement that we believe represent radiation necrosis. Tissue confirmation of necrosis was possible in three of these patients and was presumptive in the remainder because of their subsequent clinical course. Discussion The survival data in this series represent an improvement over our previous data and those of other series previously reported in the literature. However, associated with this improvement in survival, another important clinical problem emerged: a considerable proportion of patients relapsed with fresh brain metastases or suffered neurologic morbidity from delayed effects of radiation therapy. Patient selection is clearly an important factor in the superiority of results achieved by combinedmodality treatment in this and all other surgical series, and there is no comparable group reported that have been treated by radiation therapy alone. However, the overall median survival in our study exceeds that of the best prognostic subgroups reported by the Radiation Therapy Oncology Group (RTOG) for those patients treated by radiation al~ne.~~,~ We attribute this improvement to a combination of factors. More accurate detection of single brain metastases by CT scans now allows earlier diagnosis and prompt therapy to be instituted in most patients. Although size of brain metastases is rarely considered when reporting results of treatment, our experience suggests that small tumors are more likely to have a successful outcome. In this series, % of the patients had tumors measuring. cm in diameter, and there was a close correlation between small tumor size and relatively intact neurologic status. For this reason, survival was not analyzed separately with respect to size of the brain metastases. Accurate localization of the tumor by CT scans in the axial and coronal planes facilitated precise planning of the surgical approach, especially in those tumors that were deep or of irregular configuration. This resulted in a significant reduction of morbidity, while allowing gross total resection of the tumor in the majority. Continued innovations in neurosurgical technology, such as the ultrasonic aspirator and laser, have further contributed to reduction of morbidity after craniotomy for tumor resection over recent years. Although no dramatic decline in -day mortality was evident in the current series in comparison to our previous data, others have noted a progressive TABLE 6. Central Nervous System Relapses by Tumor Type Other Local sites Both TABLE 7. Central Nervous System Relapse Rates by Site of Tumor in Patients With Solitary Metastases Site No. of tumors Relapse Melanoma Sarcoma Miscellaneous 9 8 (%) 7 7 I (I %) (8%) (%) Melanoma Sarcoma Miscellaneous I (%)

6 No. 6 SURGERY FOR BRAIN METASTASES * Sundaresan and Galicich 87 decline in mortality for craniotomy over the recent years..i In a large series of 9 patients undergoing craniotomy for brain metastases, Takakura reported an overall mortality of.9%, but this dropped to % after 976 when CT scans were routinely used for radiologic examination and surgical planning. With such expected low mortality rates, it was possible for us to offer repeat craniotomy to 6% of our patients who developed local recurrence or other sites of brain metastases. In addition, most patients continued to receive various forms of systemic therapy, but the impact of such treatment is difficiilt to gauge. The favorable prognostic factors identified in this series are similar to those previously reported by us and others. -I As expected, the presence of systemic metastases and uncontrolled primary cancer were the major determinants of survival. With statistically adequate numbers of each type of tumor, important survival differences based on primary site were also evident. The best results were seen in patients with lung cancer, because the majority in this group had undergone curative resection of the original tumor and had no other sites of metastases. Although early onset of brain metastasis implies a poor prognosis in general, this was true in our study only for those with primary sites from the kidney, breast, and colon. In patients with lung cancer, no significant difference in survival was seen between patients with early and late onset of brain metastasis. These patients had a median survival of 8 months, and therefore constitute an important subgroup for whom aggressive therapy is Although the survival of patients with metastasis located in the posterior fossa was considerably poorer relative to those with supratentonal tumors, it is important to note the spectrum of primary tumors in this location. They included patients with colon and kidney cancer-two tumors that respond poorly to radiation therapy alone-and others who had failed radiation therapy. Effective comparison of neurologic palliation achieved by combined-modality treatment versus that achieved by radiation therapy alone is difficult. Most patients currently receive high-dose corticosteroid therapy, thus neurologic deficits often resolve completely. Cairncross and Posner noted that initial clinical improvement after radiation therapy in several major series varied from % to 9% of treated patienk6 However, the response was limited; the median duration was months and ranged from to 6 months. Although initial improvement may be comparable in patients treated by a variety of modalities, i.e., radiation, surgery, or even steroid therapy alone, important differences are likely to emerge only in those patients in whom expected survival exceeds months. These findings are supported by examining data for the best prognostic subgroups (ambulatory patients with disease limited to the central nervous system) reported by the RTOG studies. The -year neurologic control rates varied: 6% for patients with breast cancer, % in lung cancer patients, and 9% in all others. Only one third of the patients were stable without evidence of neurologic progression before death. In comparison, two thirds of the patients in our study were still improved at year, and % of survivors maintained this improvement at years. The major reason for the progressive loss in overall palliation was the appearance of new brain metastases (and leukoencephalopathy presumably related to therapy). In this regard, our experience parallels that of Looper et al. who have noted a high incidence of severe neurologic problems following combined-modality therapy (chemotherapy and WBRT) in long-term survivors with small cell lung cancer. The majority of patients with brain metastases die of progressive systemic cancer while neurologically Only in patients with solitary lesions will effective central nervous system therapy translate more directly into overall improvement in survival. However, this group of patients had a much higher central nervous system relapse rate in our series, because this is a reflection of prolonged survival and, consequently, a longer period of risk for relapse. Local recurrences at the original tumor site in this study were related to several factors: incomplete surgical excision; infiltrative nature of some histologic types, such as epidermoid and colon carcinoma; as well as previous therapy, i.e., radiation. In selected patients at high risk for local recurrence, additional local control might best be sought by boosting radiation doses to the tumor bed while keeping that to the whole brain to an acceptable minimum. More recently, some investigators have shown that chemotherapy, either by the intra-arterial or systemic route, has considerable value in the treatment of brain metas- tases if systemic tumor is also respon~ive.~,~~ If effective chemotherapy for solid tumors becomes available, it would offer considerable promise as adjuvant therapy in the management of patients with brain metastases after tumor resection. REFERENCES. Takakura K, Sano K, Hojo S, Hirano A. Metastatic Tumors of the Central Nervous System. Tokyo: Igaku-Shoin, 98.. Posner JB, Chernik NL. Intracranial metastases from systemic cancer. Adv Neurol 978; Posner JB. Brain metastases: A clinician s view. In: Weiss L, Gilbert HA, Posner JB, eds. Brain Metastases. Boston: G. K. Hall, 98; -9.. Espana P, Change P, Wiernik PH. Increased incidence of brain metastases in sarcoma patients. Cancer 98; : Cairncross JG, Kim JH, Posner JB. Radiation therapy for brain metastases. Ann Neurol 98; 7: Cairncross JG, Posner JB. The management of brain metastases. In: Walker MD, ed. Oncology of the Nervous System. Boston: Martinus NijhotT, 98; -77.

7 I88 CANCER March 98 VOI. 7. Kurup P, Reddy S, Hendrickson FR. Results of re-irradiation for cerebral metastases. Cancer 98; 6: Stortebecker TP. Metastatic tumors of the brain from a neurosurgical point of view: A follow-up study of 8 cases. J Neiiroszirg 9; ~89- I I. 9. Galicich JH, Sundaresan N. Intracranial metastases. In: Wilkins R, Rengacharry S, eds. Textbook of Neurosurgery. New York Williams and Wilkins, 98; (in press).. Sundaresan N, Galicich JH, Beattie El. Surgical treatment of brain metastases from lung cancer. J Neurosurg 98; 8: Galicich JH, Sundaresan N, Arbit E, Passe S. Surgical treatment of single brain metastases: Factors associated with survival. Cancer 98 : Zimm S, Wampler GL, Stablein D. lntracerebral metastases in solid tumor patients: Natural history and results of treatment. Cancer 98 I ; White KT, Fleming TR, Laws ER. Single metastasis to the brain: Surgical treatment in consecutive patients. Aayo Clin Proc 98 I; 6:-8.. Winston KR, Walsh JW, Fischer EL. Results of operative treatment of intracranial metastatic tumors. Cancer 98; : Di Stefan A, Yap HY, Hortobagyi LN et at. The natural history of breast cancer patients with brain metastases. Cancer 979; : Patchell RA, Posner JB. Comparison of treatments for single brain metastases from non-small cell lung cancer: Surgery plus radiation therapy vs. radiation therapy alone (Abstr). Neurology 98; :7. 7. Cascino TL, Leavengood JM, Kemeny N, Posner JB. Brain metastases from colon cancer. J Neurosurg 98; : Decker DA, Decker VL, Herskovic A, Cummings GD. Brain metastases in patients with renal cell carcinoma: Prognosis and treatment. J Clin Oncol 98; : Fell DA, Leavens ME, McBride CM. Surgical vs. non-surgical management of metastatic melanoma of the brain. Neurosurgery 98; 7:8-.. Byrne TN, Cascino TL. Posner JB. Brain metastasis from melanoma. J Neurooncol 98; :-7.. Pezner RD, Lipsett JA, Archambeau JD, Fine RM, Moss WT. High dose fractionated radiation therapy for select patients with brain metastases. Radiology 98 I; :79-8 I.. Borgett B, Gelber R, Kramer S et al. The palliation of brain metastases: Final results of the first two studies by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 98; 6: -9.. Gelber RD, Larson M, Borgelt BB, Kramer S. Equivalence of radiation schedules for the palliative treatment of brain metastases in patients with favourable prognosis. Cancer 98; 8: Dosoretz DE, Blitzer PH, Russell AH, Wang CC. Management of solitary metastasis to the brain: The role of elective brain irradiation following complete surgical resection. Int J Radiat Oncol Biol Phys 98; 6: Sundaresan N, Galicich JH, Tomita T, Deck MDF. Radiation necrosis after treatment of solitary intracranial metastases. Neurosurgery 98; 8: Sundaresan N, Galicich JH. Surgical treatment of brain metastases: Clinical and CT analyses of CNS failure rates (Abstr). Proc Am Soc Clin Oncol 98; C-6, Galicich JH, Sundaresan N, Thaler HT. Surgical treatment of single brain metastases: Evaluation of results by computerized tomography scanning. J Neurosurg 98; : Dixon WJ, Brown MB, eds. BMDP-79: Biomedical Computer Programs, P-series. Berkeley, California: University of California Press, Sundaresan N, Galicich JH. The surgical treatment of single brain metastasis in non-small cell lung cancer. Cancer Investigation (in press).. Looper JD, Einhom LH, Garcia SA et al. Severe neurologic problems following successful therapy for small cell lung cancer (Abstr). Proc Am Soc Clin Oncol 98; C-9:.. Cascino TL, Byrne TN, Deck MDF, Posner JB. Intra-arterial BCNU in the treatment of metastatic tumors. J Neurooncol 98; : -8.. Rosner D, Nemoto T. Pickren J, Lane W. Management of brain metastases from breast cancer by combination chemotherapy. J Neurooncol 98; :-7. American Cancer Society National Conference on Advances in the Care of the Child With Cancer This Amencan Cancer Society National Conference is to be held on June -, 98 at the Los Angeles Hilton, Los Angeles. California. Topics included in the conference are as follows: The Cure of Childhood Cancers, The Importance of Evaluation and Staging of Children With Cancer, Integrated Management During the Initial Therapy and the Acute Phase, The Long- Term Needs for Integrated Management, The Recent Developments for Future Projections in Biologic Understanding of Pediatric Cancer and Recent Developments and Future Projections for the Therapy of Pediatric Cancer. This continuing medical education activity meets the criteria for % hours in Category I of the Physicians Recognition Award of the American Medical Association. For additional information contact: American Cancer Society National Conference on Advances in the Care of the Child With Cancer, 777 Third Avenue, New York, NY 7.