2548 Prognostic Factors for Survival in Patients with Brain Metastases from Renal Cell Carcinoma Stéphane Culine, M.D., Ph.D.1 Mohamed Bekradda, M.D.1 Andrew Kramar, Ph.D.2 Annie Rey, B.Sc.2 Bernard Escudier, M.D.1 Jean-Pierre Droz, M.D.3 1 Department of Medicine, Institut Gustave Roussy, Villejuif, France. 2 Department of Biostatistics, Institut Gustave Roussy, Villejuif, France. 3 Department of Medicine, Centre Léon Bérard, Lyon, France. BACKGROUND. Patients presenting with brain metastases from renal cell carcinoma portend a poor prognosis, with a reported median survival of 4 – 6 months. Given their short life expectancy, these patients generally have been excluded from clinical trials that assess the efficacy of medical treatments. However, clinical impression suggests that some patients may achieve long term palliation. METHODS. The clinical features of 68 patients who were treated at the Institut Gustave Roussy for brain metastases from renal cell carcinoma were collected retrospectively. Using univariate and multivariate analyses, a prognostic model based on independent prognostic factors was established. An external data set of 57 patients was used to validate the model. RESULTS. The median survival was 7 months. On univariate analysis survival was related significantly to the following adverse prognostic factors: no initial nephrectomy, left side and temporal location of brain metastases, presence of fever or weight loss, erythrocyte sedimentation rate ⬎ 50 mm/h, and time from initial diagnosis to brain metastases ⱕ 18 months. Multivariate analyses identified the previous variable as well as the presence of other visceral metastases as independent prognostic factors. Forty-four patients (65%) with no or 1 adverse prognostic factor (average risk group) had a median survival of 8 months and a 26% 1-year survival rate. Twenty-four patients (35%) with 2 adverse prognostic factors (poor risk group) had a median survival of 3 months and a 1-year survival rate of 9%. This model proved to be discriminant in an external data set; the median survival of patients assigned to the average risk group was 11 months (46% 1-year survival rate) compared with 4 months (9% 1-year survival rate) for patients assigned to the poor risk group. CONCLUSIONS. Patients presenting with brain metastases from renal cell carcinoma and poor risk prognostic factors are highly unlikely to benefit from medical treatments except symptomatic procedures. Conversely, the enrollment of patients with average risk prognostic factors into clinical trials dealing with chemotherapy or immunotherapy may be considered. Cancer 1998;83:2548 –53. © 1998 American Cancer Society. KEYWORDS: brain metastases, prognostic factors, renal cell carcinoma, risk, survival. A The authors thank S. Négrier, M.D., A. Ravaud, M.D., and G. Chvetzoff for providing the clinical charts of patients included in the external data set. Address for reprints: Stephane Culine, M.D., Ph.D., Department of Medicine, CRLC Val d’Aurelle, 34298 - Montpellier Cedex 5, France. Received April 8, 1998; accepted May 7, 1998. © 1998 American Cancer Society median survival ranging from 6 –12 months has been reported in patients with metastatic renal cell carcinoma (RCC).1– 4 Patients presenting with brain metastases from RCC specifically portend an even poorer prognosis, with a median survival ranging from 17 weeks–7 months.5,6 Given their short life expectancy, these patients generally have been excluded from clinical trials that assess the efficacy of medical treatments (i.e., immunotherapy or chemotherapy). However, clinical impression suggests that some patients may achieve long term palliation. The current study presents the long term outcome of 68 patients Brain Metastases in Renal Cell Carcinoma/Culine et al. with brain metastases from RCC who were treated at the Institut Gustave Roussy over a 17-year period. We specifically investigated whether baseline clinical characteristics could be of prognostic value in predicting survival with the aim of selecting a subgroup of patients who could be included into clinical trials dealing with chemotherapy or immunotherapy. PATIENTS AND METHODS The medical records of 1235 patients with RCC who were registered at the Institut Gustave Roussy between January 1975 and June 1993 were reviewed retrospectively. The clinical charts from 68 patients (5.5%) with histologically proven RCC, confirmatory diagnostic studies (i.e., isotope brain scan, computerized tomography [CT] scan, or magnetic resonance imaging) for brain metastases, and adequate follow-up were selected. Using such criteria, only six patients who were registered for therapeutic treatment but were neither treated nor followed in the institution were eliminated from the current study. As a rule, brain CT scans were performed routinely for pretherapy staging at the time of the first metastatic event and subsequently performed during the course of disease if symptoms were present. Patient characteristics at the time of initial diagnosis of RCC are depicted in Table 1. Subsequent data extracted from the charts included disease free interval (DFI) (defined as the time from diagnosis of RCC to the diagnosis of brain metastases), performance status (PS), constitutional and neurologic symptoms, anatomic locations, erythrocyte sedimentation rate (ESR), and other concomitant metastases at the time of diagnosis of brain metastases; treatment of brain metastases; and survival (defined as the time from diagnosis of brain metastases to death or last follow-up visit). Survival was evaluated according to the Kaplan– Meier method and differences between the survival curves were assessed by the log rank test.7 Independent prognostic factors were identified by the Cox regression analysis according to a forward stepwise procedure.8 The statistical analysis was performed using STATA Statistical Software (Stata Corporation, College Station, TX). The clinical charts of patients who entered the external data set were collected in the Centre Léon Bérard (Lyon, France) and the Institut Bergonié (Bordeaux, France) using criteria similar to those described for the Institut Gustave Roussy. RESULTS Patient Characteristics at Diagnosis of Brain Metastases The characteristics of 68 patients at the time of diagnosis of brain metastases are shown in Table 2. The 2549 TABLE 1 Patient Characteristics at the Time of Initial Diagnosis of Renal Cell Carcinoma Characteristics Gender Male Female Age (yrs) Median Range Location of primary tumor Right side Left side Bilateral Initial nephrectomy No Yes Perioperative radiotherapy (N ⫽ 59) No Yes TNM staging (N ⫽ 56) T1 T2 T3a T3b T3c T4 N0 N1 N2 N3 M0 M1 Robson stage (N ⫽ 66) I II III IV No. 50 18 53 15–73 30 36 2 9 59 54 5 2 14 17 14 7 2 41 7 7 1 40 16 2 15 20 29 median age was 56 years (range, 21–74 years). The median DFI was 18 months (range, 0 –142 months). Brain metastases were present at diagnosis of RCC in seven patients, four of whom had neurologic symptoms. The brain was the first metastatic location in eight nephrectomized patients. In 53 patients brain metastases occurred during the course of a previously known metastatic disease. Metastases that were diagnosed before brain involvement were located in the lung (40 patients), bone (16 patients), and liver (6 patients). The diagnosis of brain metastases was performed on systematic brain CT scan in 13 patients without any neurologic symptoms. Neurologic signs were noted in 55 patients and varied depending on the location and extent of brain metastases (Table 2). The PS was equal to 0 or 1 in 25 patients, 2 in 30 patients, and 3 or 4 in 13 patients. Weight loss of ⬎ 10% of total body weight was observed in 49 patients. Ten patients 2550 CANCER December 15, 1998 / Volume 83 / Number 12 TABLE 2 Patient Characteristics at the Time of Diagnosis of Brain Metastases Characteristics Age (yrs) Median Range Disease free interval (mos) Median Range Performance status 0 or 1 2 3 or 4 Neurologic symptoms None Motor weakness Intracranial hypertension Cranial nerve palsy Confusion/disorientation Cerebellar ataxia Seizure Aphasia Sensitive deficiency Weight loss No Yes Fever No Yes Erythrocyte sedimentation rate (N ⫽ 42) Median Range Treatment Surgery Radiotherapy Immunotherapy Chemotherapy None TABLE 3 Distribution of Supratentorial Metastases No. (%) 56 21–74 18 0–142 Side Location Left Right Total Frontal Parietal Temporal Occipital Total 13 25 12 14 64 11 9 3 4 27 24 34 15 18 91 25 (37) 30 (44) 13 (19) 13 (19) 22 (32) 20 (29) 15 (22) 13 (19) 8 (12) 7 (10) 6 (9) 3 (4) 19 (28) 49 (72) 58 (85) 10 (15) 50 (2–130) 10 57 14 13 8 volved. Conversely, in patients with multiple brain metastases, 54 metastases were identified in the left side of the brain, mostly in the parietal area. Treatment of Brain Metastases Eight patients received no specific treatment. Ten patients underwent complete surgical resection of a solitary metastasis; seven of these ten patients received postoperative radiation therapy. Radiotherapy was delivered to the whole brain by parallel opposed lateral fields in fifty-seven patients. The tumor doses were 18 gray (Gy) in 3 fractions (31 patients); 36 Gy in 6 fractions (8 patients); 30 Gy in 12 fractions (15 patients); and 30 Gy in 5 fractions (3 patients). A significant symptomatic improvement was observed in 38 of 50 patients who received radiotherapy alone. However, no objective response was observed among the 32 patients who underwent a postirradiation brain CT scan. Additional symptomatic therapy included corticosteroids in 61 patients and antiepileptic drugs in 43 patients. Prognostic Factors for Survival presented with fever. The ESR was noted in 42 patients and the median value was 50 mm/h (range, 2–130). Characteristics of Brain Metastases All 68 patients had confirmed brain metastases on CT scan (63 patients), magnetic resonance imaging (2 patients), or isotopic brain scan (3 patients). The number and location of brain metastases could be described accurately in 66 patients. A total number of 104 brain metastases were identified, 13 of which were located in the infratentorial area. Among the 91 supratentorial metastases, 64 (70%) were located in the left side of the brain and the parietal area was predominantly involved (Table 3). Thirty-three patients (50%) had a solitary metastasis, which in 3 cases was a unique infratentorial metastasis. Twenty metastases were located in the right side of the brain and the frontal and parietal areas were predominantly in- The median survival for all 68 patients was 7 months (Fig. 1). Five patients were still alive 6, 6, 17, 23, and 61 months, respectively, after the diagnosis of brain metastasis. The median survival of the 8 patients who received no specific therapy was 1 month. The median survival of the patients who underwent surgery (10 months) was statistically better than the median survival of the patients treated by radiotherapy alone (7 months) (P ⫽ 0.04). Using P ⬍ 0.1 as a cutoff point in univariate analysis, survival was related significantly to the following pretreatment adverse prognostic factors: no initial nephrectomy, left-sided brain metastases, temporal location, presence of fever or weight loss, ESR ⬎ 50, and time from initial diagnosis to brain metastases ⱕ 18 months (Table 4). PS was not statistically significant mainly due to the PS 2 category, which followed a survival pattern similar to the PS 3 category during the Brain Metastases in Renal Cell Carcinoma/Culine et al. 2551 TABLE 4 Univariate Analysis of Prognostic Factors Variable FIGURE 1. Overall survival of the 68 patients with brain metastases from renal cell carcinoma. first 3 months and joined the survival pattern of the PS 0/1 category after 9 months. This fact led to a univariate nonsignificant statistical result due to nonproportional hazard rates. The multivariate analysis first was performed using all variables found to be significant in the univariate analysis except ESR, which was omitted because it was only available in 42 patients. The other variables, which were not significant in the univariate analysis, then were added to the model to assess how they were likely to contribute significant information when adjusted for the factors in the multivariate analysis. Two independent adverse prognostic parameters were retained: the presence of other metastases (P ⫽ 0.017) and a DFI ⱕ 18 months (P ⫽ 0.002). A classification into average risk or poor risk groups then was made depending on the number of adverse prognostic factors (Table 5). Forty-four patients (65%) with no or only 1 adverse prognostic factor (average risk group) had a median survival of 8 months and a 1-year survival rate of 26%. Twenty-four patients (35%) with 2 adverse prognostic factors (poor risk group) had a median survival of 3 months and a 1-year survival rate of 9% (Fig. 2). This prognostic model proved to be discriminant in an external data set of 57 patients treated in 2 independent cancer centers. In general patient characteristics did not differ significantly from those used in obtaining the prognostic groups. However, there were significantly more patients assigned to the poor risk group (60%) because of a higher proportion of patients with a short DFI. The median survival of patients assigned to the average risk group was 11 months (1-year survival rate of 46%) compared with 4 months for patients assigned to the poor risk group (1-year survival rate of 9%). Side of primary tumor Right Left Bilateral Initial nephrectomy No Yes Robson stage I–II III IV Number of brain metastases 1 ⬎1 Right side No Yes Left side No Yes Frontal location No Yes Parietal location No Yes Temporal location No Yes Occipital Location No Yes Infratentorial location No Yes Performance status 0/1 2 3 Fever No Yes Weight loss No Yes ESR ⱕ50 ⬎50 Other metastases No Yes Disease free interval (mos) ⱕ18 ⬎18 ESR: Erythrocyte sedimentation rate. No. of patients Median survival (mos) 30 36 2 7 4 4 9 59 3 7 17 20 29 5 8 4 33 33 7 7 0.3 14 52 3 7 0.1 27 39 9 5 0.04 41 24 7 5 0.5 31 34 6 7 0.2 50 15 7 6 0.06 47 18 7 7 0.8 52 13 6 7 0.4 25 30 13 8 4 3 58 10 7 3 0.03 19 49 9 4 0.07 25 17 8 4 0.09 15 53 9 4 0.16 34 34 4 9 0.01 P value 0.9 0.07 0.8 0.2 2552 CANCER December 15, 1998 / Volume 83 / Number 12 TABLE 5 Prognostic Model for Survival Prognostic group Model set Average risk Poor risk Validation set Average risk Poor risk No. of adverse prognostic variables No. of patients (%) Median survival (mos) 0 or 1 2 44 (65%) 24 (35%) 8 3 0 or 1 2 23 (40%) 34 (60%) 11 4 FIGURE 2. Survival curves according to the prognostic model. Solid line: average risk group; dashed line: poor risk group. DISCUSSION Of 1235 patients with RCC registered at Institut Gustave Roussy over a 17-year period, we selected 68 patients (5.5%) who presented with brain metastases. This incidence is similar to that reported in another clinical series.5 However, autopsy studies have noted a larger incidence of brain metastases from RCC ranging from 10 –20%.9,10 These discrepancies may result from the observation that many patients have clinically unrecognized intracranial disease and often experience greater morbidity and mortality related to other visceral metastases.11 Indeed, brain metastases were diagnosed fortuitously on CT scans in 13 patients who had no neurologic symptoms. Conversely, symptomatic brain metastases caused four patients with no previously diagnosed primary neoplasm to seek medical attention. The kidney was reported to be the site of primary tumor in 5–10% of patients presenting with initial brain metastases.12–14 In another eight patients, the brain was the first metastatic location. Although rather unusual in RCC, such a dissemination pathway without any other evidence of disease has been described previously.15,16 With regard to therapeutic approaches, corticosteroids should be the first intervention for every pa- tient with brain metastases.17 Although their exact biochemical mechanisms are not known, they reduce peritumoral edema, thus diminishing the total mass of brain metastases as well as local brain compression and reduce neurologic signs.18 However, corticosteroids customarily are contraindicated during the course of immunotherapy because they may abrogate the immune response. Therefore, patients should discontinue steroid use to be eligible for potential treatment with immunotherapy. This means that the brain metastases do not progress after local treatment (i.e., surgery and/or radiotherapy). In our series local therapies included surgery in 10 patients with a solitary metastasis and external beam whole brain irradiation alone in 50 patients. Although the objective of the current study was not to compare the results of treatment, we observed a significantly better survival in favor of surgery in univariate analysis. However, this prognostic significance was not retained in the multivariate analysis. Two prospective randomized studies have suggested the superiority of surgery plus radiotherapy over radiotherapy alone in patients with a single brain metastasis.19,20 However a third trial failed to confirm these results.21 Although these studies were not restricted to patients with brain metastases from RCC, it appears appropriate that a patient with a single brain metastasis should be considered for surgical resection as the primary treatment, especially young patients with controlled or absent extracranial tumor activity.22 Cranial irradiation remains the mainstay of therapy when surgery is contraindicated. The majority of patients in the current study experienced neurologic improvement but no objective response was observed. These results are in agreement with other series in which RCC usually was unresponsive to conventional photon therapy and only a subjective improvement was observed.23,24 Recent publications have focused on prognostic factors for survival in patients with metastatic RCC who were treated by chemotherapy,3,4 interleukin-2, or interferon-␣.25–27 Multivariate analyses identified PS, weight loss, sarcomatoid histology, number of metastatic sites, time from initial diagnosis to systemic therapy, and ESR as strong predictors for survival. To our knowledge, no multivariate analysis has been published so far in patients with brain metastases from RCC. Only one previous study focused on variables influencing survival in those patients.6 PS was the most significant prognosticator in this univariate analysis performed on 34 patients.6 In our series of 68 patients, the median survival was 7 months. This observation is similar to that published by other re- Brain Metastases in Renal Cell Carcinoma/Culine et al. searchers.5,6 Systemic symptoms such as fever and weight loss only were selected as prognostic variables in the univariate analysis. The multivariate analysis retained the presence of other visceral metastases and a short DFI (ⱕ 18 months) as adverse prognostic factors. The ESR also retained a significant prognostic value but was omitted from the prognostic model because it was available in only a small number of patients. Using our prognostic model, which was validated in an external data set of patients, a poor risk subgroup of patients with two adverse variables who are highly unlikely to benefit from medical treatments except symptomatic procedures can be distinguished. Conversely, for these patients (assigned into the average risk group) the median survival was 8 months, which is not significantly different from the median survival of patients with other metastatic sites from RCC. We suggest that patients with such characteristics are likely to benefit from optimal treatment plans. They should be treated primarily with surgery and/or radiotherapy. Subsequently, provided that brain metastases are not progressive after local treatment(s), enrollment in trials dealing with chemotherapy or immunotherapy may be considered. 12. 13. 14. 15. 16. 17. 18. 19. 20. REFERENCES 1. Maldazys JD, DeKernion JB. Prognostic factors in metastatic renal carcinoma. J Urol 1986;136:376 –9. 2. Golimbu M, Joshi P, Sperber A, Tessler A, Al-Asraki J, Morales P. Renal cell carcinoma: survival and prognostic factors. Urology 1986;27:291–7. 3. De Forges A, Rey A, Klink M, Ghosn M, Kramar A, Droz JP. Prognostic factors of adult metastatic renal carcinoma: a multivariate analysis. Semin Surg Oncol 1988;4:149 –54. 4. Elson PJ, Witte RS, Trump DL. Prognostic factors for survival in patients with recurrent or metastatic renal cell carcinoma. Cancer Res 1988;48:7310 –3. 5. Gay PC, Litchy WJ, Cascino TL. Brain metastasis in hypernephroma. J Neurooncol 1987;5:51– 6. 6. Decker DA, Decker VL, Kerskovic A, Cummings GD. Brain metastases in patients with renal cell carcinoma: prognosis and treatment. J Clin Oncol 1984;2:169 –73. 7. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:467–71. 8. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50:163–7. 9. Posner JB, Chernick NL. Intracranial metastases from systemic cancer. Adv Neurol 1978;19:579 –91. 10. Saitoh H. Distant metastases of renal carcinoma. Cancer 1981;48:1487–91. 11. Saitoh H, Hida M, Nakamura K, Shimbo T, Shiramiz T, Satoh 21. 22. 23. 24. 25. 26. 27. 2553 T. Metastatic processes and a potential indication of treatment for metastatic lesions of renal adenocarcinoma. J Urol 1982;128:916 – 8. Ebels EJ, Van Der Meulen JD. Cerebral metastases without known primary tumour: a retrospective study. Clin Neurol Neurosurg 1978;80:195–7. Le Chevalier T, Smith FP, Caile P, Constants JP, Rouesse JG. Sites of primary malignancies in patients presenting with cerebral metastases. Cancer 1985;56:880 –2. Merchut MP. Brain metastases from undiagnosed systemic neoplasms. Arch Intern Med 1989;149:1076 – 80. Taxy JB. Renal adenocarcinoma presenting as a solitary metastasis: contribution of electron microscopy to diagnosis. Cancer 1981;48:2056 – 62. Ammirati M, Samii M, Skaf G, Sephernia A. Solitary brain metastasis 13 years after removal of renal adenocarcinoma. J Neurooncol 1993;15:87–90. Ruderman NB, Hall TC. Use of glucocorticoides in the palliative treatment of metastatic brain tumors. Cancer 1965; 18:298 –306. Bruce JN, Criscuols GR, Merrill MJ, Moquin RR, Blacklock JB, Oldfield EH. Vascular permeability induced by protein product of malignant brain tumors: inhibition of dexamethasone. J Neurosurg 1987;67:880 – 4. Patchell RA, Tibbs PA, Walsh JW, Dempsey RJ, Maruyama Y, Kryscio RJ, et al. A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 1990; 322:494 –500. Vecht CJ, Haaxma-Reiche H, Noordijk EM, Padberg GW, Voormolen JH, Hoekstra FH, et al. Treatment of single brain metastasis: radiotherapy alone or combined with neurosurgery? Ann Neurol 1993;33:583–90. Mintz AH, Kestle J, Rathbone MP, Gaspar L, Hugenholtz H, Fisher B, et al. A randomized trial to assess the efficacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 1996;78:1470 – 6. Noordijk EM, Vecht CJ, Haaxma-Reiche H, Padberg GW, Voormolen JHC, Hoekstra FH, et al. The choice of treatment of single brain metastasis should be based on extracranial tumor activity and age. Int J Radiat Oncol Biol Phys 1994; 29:711–7. Omfrey V, Mohiuddin M. Radiation therapy in the treatment of metastatic renal cell carcinoma. Int J Radiat Oncol Biol Phys 1985;11:2007–9. Maor MH, Frias ARE, Oswald MJ. Palliative radiotherapy for brain metastases in renal carcinoma. Cancer 1988;62: 1912–7. Palmer PA, Vinke J, Philip T, Négrier S, Atzpodien J, Kirchner H, et al. Prognostic factors for survival in patients with advanced renal cell carcinoma treated with recombinant interleukin-2. Ann Oncol 1992;3:475– 80. Fossa SD, Kramar A, Droz JP. Prognostic factors and survival in patients with chemotherapy of interferon-␣. Eur J Cancer 1993;30A:1310 – 4. Mani S, Todd MB, Katz K, Poo WJ. Prognostic factors for survival in patients with metastatic renal cancer treated with biological response modifiers. J Urol 1995;154:35– 40.