Time From Treatment to Subsequent Diagnosis of Brain Metastases in Stage III Non每Small-Cell Lung Cancer: A Retrospective Review by the South
http://www.100md.com
▲還散笫雖悝◎
the University of Colorado Health Sciences Center, Denver CO
Loyola University Medical Center, Maywood IL
University of Washington Medical Center, Seattle WA
ABSTRACT
PURPOSE: A retrospective review of the Southwest Oncology Group (SWOG) database was undertaken to review the incidence and timing of diagnosis of brain metastases in patients undergoing combined-modality therapy for stage III non每small-cell lung cancer (NSCLC).
PATIENTS AND METHODS: Four hundred twenty-two eligible, assessable patients with stage IIIA/B NSCLC were treated on four SWOG protocols. Treatment varied with protocol but consisted of concurrent cisplatin-etoposide and radiation in all patients, with a surgery arm in two of the four protocols.
RESULTS: Of the 422 total patients, 268 (64%) have experienced disease progression; 54 relapses (20%) were in brain only, 17 (6.5%) were in brain and other sites simultaneously, and 197 (63.5%) were in sites other than brain. Of the 268 patients with disease progression, progression in the brain only, in the brain and other sites, and not in the brain occurred in 20%, 6%, and 74% of patients, respectively. Time from treatment to diagnosis of disease progression in the brain in 71 patients was as follows: during treatment, 16 relapses (22.5%); 0 to 16 weeks after treatment, 17 relapses (24%); 16 weeks to 6 months after treatment, 10 relapses (14%); 6 to 12 months after treatment, 16 relapses (22.5%); and more than 12 months after treatment, 12 relapses (17%). Nonsquamous histology and young patient age were the only significant predictors for increased risk of early relapse with brain metastases.
CONCLUSION: Brain metastases often develop early in the course of treatment for stage IIIA/B NSCLC. The statistical designs of ongoing trials of prophylactic cranial irradiation in stage III NSCLC have taken this into account.
INTRODUCTION
Brain metastases from non每small-cell lung cancer (NSCLC) have gained attention recently because combined-modality therapy has lead to improvements in local control and overall survival rates. With improved survival after combined-modality therapy, it has been noted that the incidence of brain metastases, often as a sole site of relapse, has increased.1每3 Andre et al,2 in a retrospective review of clinical N2 patients treated with surgery, found that the cumulative incidence of brain metastases in patients treated with or without neoadjuvant chemotherapy was 32% and 18%, respectively (P < .05). Ceresoli et al3 determined that, in stage III NSCLC patients treated with combined-modality therapy, the brain was the site of first relapse in 23% of patients and, ultimately, brain metastases developed in 50% of patients at some point in the course of their disease. This retrospective review was undertaken to determine the incidence of brain metastases, as well as predictive and prognostic factors for brain metastases in four Southwest Oncology Group (SWOG) studies of combined-modality therapy in stage III NSCLC.
PATIENTS AND METHODS
A retrospective review was undertaken of four SWOG studies in stage III NSCLC.4每7 The minimum follow-up time on these studies was 4 years. All studies required signed informed consent.
SWOG 8805 was a phase II study to assess the feasibility of and outcomes from concurrent chemotherapy and irradiation followed by surgery in pathologically proven stage IIIAN2 or IIIB NSCLC.4 The eligibility criteria included a performance status of 0, 1, or 2 and a forced expiratory volume in 1 second of 2 L or a predicted postresection forced expiratory volume in 1 second of at lease 1 L. Patients initially received two cycles of cisplatin (50 mg/m2 on days 1, 8, 29, and 36) and etoposide (50 mg/m2 on days 1 through 5 and 29 through 33) with concurrent chest radiotherapy to 45 Gy that began on day 1. Surgical resection was attempted if there was a response or stable disease. Postoperative chemoradiotherapy, consisting of two more cycles of cisplatin-etoposide with completion of radiation to total of 59.4 Gy, was administered if there was either unresectable disease or positive margins or nodes found at the time of surgery. The protocol allowed prophylactic cranial irradiation (PCI; 30 Gy in 15 fractions, 2 Gy per fraction) to be administered at the option of the investigator. To be eligible at study entry, patients had to have a negative computed tomography (CT) scan or magnetic resonance imaging (MRI) of the brain within 30 days of registration. The re-evaluation before surgery also required a CT or MRI of the brain. CT or MRI of the brain was performed at 6, 12, and 18 months, and then yearly for 3 years. The study met its accrual goal in March 1992. One hundred twenty-six patients participated in the study; 60% had stage IIIAN2 disease, and 40% had stage IIIB disease. The 18 patients who received PCI are excluded from this analysis. The median survivals for patients with stage IIIA and IIIB disease were 13 and 17 months, respectively. The 3-year survival rates for patients with stage IIIA and IIIB disease were 27% and 24%, respectively.
SWOG 9019 was initially designed as a phase III study in patients with pathologic stages IIIAN2 and IIIB NSCLC. Eligibility criteria were identical to SWOG 8805, as discussed earlier. Eligible patients had to have a normal CT or MRI of the brain within 30 days of study registration. Patients were randomly assigned to either arm 1 (two cycles of cisplatin 50 mg/m2 on days 1, 8, 29, and 36 and etoposide 50 mg/m2 on days 1 through 5 and 29 through 33 with concurrent thoracic radiation to 61 Gy) or arm 2, which was the identical trimodality therapy as administered in SWOG 8805. All patients received two more cycles of cisplatin and etoposide using the same dose and schedule as used during the concurrent portion of treatment. CT or MRI of the brain was required before consolidation chemotherapy only if new symptoms developed. A CT or MRI of the brain was required at 12, 18, and 24 months of follow-up, and then yearly thereafter. Shortly into its accrual, the enrollment of all patients with stage IIIAN2 disease onto both arms of patients with stage IIIB disease onto the surgical arm was discontinued to support one North American trial addressing the surgical question (INT 0139). Thus, SWOG 9019 became a single-arm study of definitive chemoradiotherapy for patients with stage IIIB disease, and accrual was completed and reported as such.6 There were 50 eligible patients with the following tumor substages: 18 patients, T4N0/1; 12 patients, T4N2; and 20 patients, N3. The overall median survival time was 15 months, and the 3-year survival rate was 17%. For this report on brain metastases, however, all of the originally enrolled patients (both arms, both stages) are included because mature follow-up and site of relapse data are available.
SWOG 9416 was a phase II study in patients with NSCLC of the superior sulcus.7 This trial tested the feasibility of induction chemoradiotherapy and surgical resection in NSCLC of the superior sulcus. Eligible patients required a normal CT or MRI of the brain within 42 days of registration. Patients with T3-4N0-1 superior sulcus NSCLC received two cycles of cisplatin (50 mg/m2 on days 1, 8, 29, and 36) with etoposide (50 mg/m2 on days 1 through 5 and 29 through 33). This chemotherapy was concurrent with 45 Gy of thoracic radiation therapy. Patients with stable or responding disease underwent thoracotomy 3 to 5 weeks later. A CT or MRI of the brain was required before thoracotomy. Two more cycles of cisplatin-etoposide were to be administered postoperatively. The study was closed in August 1999. The 2-year survival rate was 55% for all eligible patients and 70% for patients who underwent a complete resection.
SWOG 9504 was a phase II study in patients with pathologically proven stage IIIB NSCLC that was designed to test the concept of taxane sequencing in combined-modality therapy. The eligibility criteria were identical to SWOG 9019. Eligible patients had to have a CT or MRI of the brain documenting no brain metastases within 42 days of registration. Patients received cisplatin (50 mg/m2 on days 1, 8, 29, and 36) and etoposide (50 mg/m2 on days 1 through 5 and 29 through 33). This chemotherapy was administered with concurrent thoracic radiation to a dose of 61 Gy, given in 1.8-Gy daily fractions to 45 Gy and then 2.0-Gy daily fractions for an additional 16 Gy. Consolidation docetaxel started 4 to 6 weeks after completion of concurrent chemoradiotherapy. A CT or MRI of the brain was required before consolidation chemotherapy only if there were new symptoms. The docetaxel dose started at 75 mg/m2 but was to be increased to 100 mg/m2 in cycles 2 and 3 if tolerated.5 During follow-up, a CT or MRI of the brain was required at 12, 18, and 24 months, and then yearly thereafter. SWOG 9504 closed in December 1998 after having accrued 83 eligible patients with the following stage subsets: T4N0-1, 31 patients (37%); T4N2, 22 patients (27%), and T1-3N3, 30 patients (36%). The median progression-free survival time was 16 months. The survival rates at 1, 2, and 3 years were 76%, 54%, and 37%, respectively.
A statistical analysis was performed to look at the incidence and timing of development of brain metastases in the 422 eligible, assessable patients enrolled onto the four protocols, with the exception of the 18 patients on SWOG 8805 who were known to have received PCI. The analysis includes patients for whom progression was never observed or recorded. For those patients who progressed but for whom the site of progression was not recorded, often because the patients died of their disease before restaging, it is assumed that progression was at a site other than the brain. Logistic regression was performed with the outcome being brain metastases (yes or no). A Cox regression analysis, with the outcome being time to brain metastases (with censoring for all other outcomes including progression at a site other than brain), was also performed. Failure during treatment includes failures that occurred during any phase of treatment (ie, concurrent chemoradiotherapy or surgery or consolidation chemotherapy). Only the site(s) of first progression was recorded. The following variables were entered into univariate analyses as well as a proportional hazards model with stepwise selection: sex, stage (IIIA v IIIB), performance status (Eastern Cooperative Oncology Group performance status of 0 v 1 v 2), histology, planned docetaxel consolidation, age, and planned surgery (yes or no). Estimates for the risk of brain relapse were generated using cumulative incidence with competing risks for death or progression to other sites.
RESULTS
There were 422 eligible patients on the four SWOG studies, of whom 64% have had documented tumor progression or relapse. Table 1 lists the total number of patients per study and the percentage of patients in each study in whom progression was recorded. Progression was categorized as brain only (brain metastases with no other site of failure), brain and other sites (brain metastases with simultaneous failure in extracranial site), or not brain (extracranial failure only). Progression of the primary lung cancer was counted as an extracranial failure. Progression expressed as a percentage of the 422 total patients was found to be brain only, brain and other sites, and not brain in 13%, 4%, and 47% of patients, respectively (Table 2). Of the 268 patients with progression, progression in the brain only, in the brain and other sites, and not in the brain occurred in 20%, 6%, and 74% of patients, respectively (Table 3).
We examined potential predictors of progression site using both univariate and stepwise logistic regression using the following variables: SWOG study, planned surgery (yes or no), planned docetaxel consolidation (yes or no), stage (IIIB v IV), age, and histology. The only variables that reached statistical significance were histology and patient age at presentation. The incidence of disease progression according to histology is summarized in Table 4. There was no significant risk of overall progression according to histology. However, nonsquamous histology was associated with a higher risk of progression within the brain, either as the sole site of relapse or concurrent with failure at other sites (P = .007, odds ratio = 2.4 for nonsquamous v squamous histology; Table 5).
Table 6 summarizes the time of first failure relative to study registration for the 268 patients who have had documented progression. This table also indicates the time of failure according to the site of progression. Seventy-eight patients (29.1% of all patients who experienced treatment failures) had documented progression during the course of their therapy. Eighty percent of these failures were at distant sites excluding the brain. Time from study registration to the diagnosis of brain metastases in the 71 patients who developed brain metastases was as follows: during treatment, 16 patients (22.5%); 0 to 16 weeks after treatment, 17 patients (24%); 16 weeks to 6 months after treatment, 10 patients (14%); 6 to 12 months after treatment, 16 patients (22.5%); and more than 12 months after treatment, 12 patients (17%). The time of first failure was also calculated from the date of study registration. Time from study registration to the diagnosis of brain metastases in the 71 patients who developed brain metastases was as follows: 0 to 6 months from registration, 26 patients (37%); 6 to 12 months from registration, 23 patients (32%); and more than 12 months from registration, 22 patients (31%).
Brain only as a site of failure was found in 54 patients (20% of all patients who experienced treatment failures), and 14% of the failures occurred during treatment. Almost half of all failures (46.5%) within the brain (brain only or brain and other sites) occurred during or within 16 weeks of completing therapy. Nonsquamous histology was associated with a shorter time to relapse in the brain compared with squamous carcinomas. Multivariable Cox regression analysis yielded results similar to the logistic regression analysis. Histology and patient age were the only significant predictors for brain metastases among the variables examined. The hazard ratio was 2.1 for nonsquamous histology versus squamous histology (P = .02). Younger patients were at increased risk for developing brain metastases. The hazard ratio was 1.8 for age 50 years versus more than 50 years (P = .046), Factors that were not found to predict time of failure were intent for surgery, sex, use of taxane consolidation, performance status, and stage (IIIA v IIIB).
The median survival times measured from the date of study registration were 26, 12, and 16 months for patients with relapse in brain only, relapse in brain and other sites, and distant failure not in brain, respectively (Fig 1). The median survival times measured from the date of progression were 10.3, 4.5, and 5 months for patients with relapse in brain only, relapse in brain and other sites, and distant failure not in brain, respectively. The median survival after progression of 10.3 months for patients with relapse in the brain only was significantly different from the 5-month survival of patients with all other sites of failure (P = .0002). The interval from study registration to tumor progression was a factor predictive of subsequent survival. Median survival time after progression for patients who experienced treatment failure within 6 months or after more than 6 months from study registration was 5 and 8 months, respectively (P = .05).
The overall 2-year cumulative incidence of brain failure was 17% (Fig 2). Patients with adenocarcinoma had the highest rate of brain failure, with a cumulative incidence of 22% v only 10% for patients with squamous cell histology (Fig 3). Patients in the 50- to 60-year age range had a comparable risk of brain failure (12% to 16% 2-year incidence), whereas patients presenting at age less than 50 years were much more likely to have the brain as their first site of failure, with a 2-year incidence of 26%.
DISCUSSION
This analysis found that, for stage III NSCLC, the factors on multivariate analysis that predicted for the subsequent development of brain metastases were histology and age less than 50 years. Patients with nonsquamous histology were more likely than patients with squamous cell histology to progress at any site and also had an increased risk of having relapse with brain metastases. The 2-year cumulative risk of failure in the brain for patients with adenocarcinoma and squamous cell carcinoma was 22% and 10%, respectively. The influence of histology on the risk of brain metastases has been observed by others.2,8,9 Ryan et al8 reported the breakdown by histology in one large series of patients with brain metastases from lung cancer as follows: 31% small-cell carcinoma, 21% adenocarcinoma, 21% large-cell carcinoma, and 8% squamous cell carcinoma. Andre et al2 reported the incidence of isolated brain metastases in patients treated with cisplatin-based neoadjuvant chemotherapy and surgery. Isolated brain metastases occurred in 6%, 28%, and 8% of patients with squamous cell carcinoma, adenocarcinoma, and other histologies, respectively (P .05). This study found the following factors not to be predictive of subsequent development of brain metastases: sex, performance status, and stage (IIIA v IIIB).
Other studies have found a range of factors to be significant. Ceresoli et al3 reported that stage III NSCLC patients were more likely to develop brain metastases if they were younger than 60 years of age. This observation is in agreement with our observation of an increased risk of brain metastases in patients less than age 50 years of age. Ceresoli et al3 also found that the bulk of mediastinal lymphadenopathy (> 2 cm v 2 cm) was a significant factor on univariate analysis but that it lost its significance on multivariate analysis. Our study was unable to address this issue of bulk of mediastinal lymphadenopathy.
The choice of treatment likely affects the pattern of tumor relapse. Andre et al2 conducted a retrospective review and concluded that the risk of brain metastases was increased in patients who received neoadjuvant chemotherapy followed by surgery. The 3-year incidence of isolated brain metastases in patients with clinical N2 NSCLC treated with or without neoadjuvant chemotherapy was 20% and 9%, respectively. Robnett et al10 found that the sequencing of chemotherapy and radiation therapy in locally advanced NSCLC was associated with significantly different rates of subsequent brain metastases. The crude and 2-year actuarial risk of brain metastases was 27% and 39%, respectively, when induction chemotherapy was followed by radiation therapy. For patients having immediate concurrent chemoradiotherapy, the crude and 2-year actuarial risk of brain metastases was 15% and 20%, respectively. Our study found that the use of taxane consolidation did not affect the subsequent risk of developing brain metastases.
Our study determined that the survival rate after diagnosis of brain metastases was significantly longer for patients with failure in the brain only as opposed to failure simultaneously in the brain and other sites. This is in keeping with other retrospective reviews, including the recursive partitioning analysis (RPA) by the Radiation Therapy Oncology Group (RTOG).11每14 The RTOG RPA examined prognostic factors in 1,200 patients with brain metastases from all primary solid tumors treated on RTOG studies.11 Sixty-one percent of those patients had primary lung cancer, with only a small number of patients with small-cell histology. Variables found to be of significant prognostic importance were performance status, extent of extracranial disease, control of primary cancer, and age. In another relatively large retrospective review of 250 patients with brain metastases from NSCLC, patients with an absent or controlled primary tumor had a 1-year survival rate of 26% as opposed to just 11% for patients with an active primary tumor (P = .051).13
This study found that patients who developed brain metastases within 6 months of study registration did significantly worse than patients who developed brain metastases at a later date (median survival, 8 v 6 months; P = .05). The prognostic impact of the time interval between the diagnosis of the primary NSCLC and brain metastases has been found in other retrospective series.9,15 Abrahams et al,15 at the Hospital of the University of Pennsylvania, reported an improved outcome for patients with brain metastases diagnosed more than 3 months from diagnosis of the primary cancer compared with patients with brain metastases diagnosed earlier. The median survival time and 5-year survival rate for metachronous versus synchronous brain metastases were 18 months and 28.9% v 9.9 months and 0%, respectively. The interval between diagnosis and developing brain metastases was also a factor predictive of survival in a series of NSCLC patients treated with radiosurgery.9 The median survival time from the date of radiosurgery was 11 months for patients diagnosed more than 7 months from diagnosis of primary cancer compared with just 7 months for patients with a shorter interval. The RTOG RPA found that the interval from diagnosis of the primary cancer to the development of brain metastases ( 2 years v > 2 years) was of no prognostic importance.11 This discrepancy between the RTOG RPA conclusions and those of other studies could have been a result of the time point chosen (ie, 2 years in the RTOG analysis v 3 to 7 months in other studies).
The most important observation from this study is that the interval between study registration and the development of brain metastases was so short. Forty-six percent of patients who developed brain metastases had these diagnosed within 16 weeks of completing treatment. The two factors on multivariate analysis that were predictive of a short interval to the development of brain metastases were nonsquamous histology and young patient age. The hazard rate for brain metastases was significantly higher in patients with adenocarcinoma or other nonsquamous histologies and significantly higher for patients who were age 50 years or younger.
There have been few reports that deal with the issue of timing of the development of brain metastases in patients with initially locally advanced NSCLC. Ceresoli et al3 reported the time to brain relapse measured from the initiation of treatment in 112 patients with stage IIB, IIIA, or IIIB NSCLC. All patients were treated with neoadjuvant chemotherapy (cisplatin, vinblastine, and mitomycin) and then re-evaluated for surgery. Patients considered to have potentially resectable disease went on to surgery and postoperative radiation. Unresectable patients were treated with radiation alone. The median time to brain relapse measured from the initiation of therapy was just 9 months. The only two factors predictive of a short time to relapse within the brain in multivariate analysis were young age (less than 60 years) and mediastinal lymphadenopathy greater than 2 cm in diameter. The median time to brain relapse in patients with squamous and nonsquamous histology was 21 and 14 months, respectively, although this difference did not reach statistical significance.
Using standard patient and tumor characteristics as predictors of relapse within the brain has obvious limitations in terms of positive predictive value. Another approach is to look at molecular markers. Investigators have looked at molecular markers and their association with the subsequent occurrence of brain metastases in NSCLC.16,17 D'Amico et al16 retrospectively reviewed 202 pathologic specimens from patients undergoing resection of stage I NSCLC, of whom 25 were subsequently found to have isolated brain metastases. A panel of eight molecular markers was chosen to represent multiple pathways in the metastatic process: growth regulation (erbB2), apoptosis (p53), angiogenesis (factor viii), cellular adhesion (EphA2 and E-cadherin), and basement membrane invasion (UPA, UPAR, and PAI-1). Patients with isolated brain relapse had significantly higher expression of p53 and UPA. In addition, patients with E-cadherin expression had a low incidence of developing brain metastases. Bubb et al17 found that Ki-67, p53, and bcl-2 in primary NSCLC primary tumors were not predictive of survival or the incidence of brain metastases. There needs to be more work done in this area.
In conclusion, this study demonstrated a high rate of early relapse within the brain in patients with stage III NSCLC treated with multimodality therapy. Many of the failures were isolated brain metastases. These results argue for some form of treatment to prevent the development of brain metastases. PCI has been proposed as one such method of reducing the incidence of brain metastases. Many cooperative groups within North America have joined the RTOG-led phase III study in which patients with stage III NSCLC are randomly assigned to either receive or not receive PCI. RTOG L-0214 was originally open to patients with stage IIIA or IIIB NSCLC who had completed all planned definitive locoregional therapy. Patients could be randomized to PCI or no PCI for up to 16 weeks after completion of all therapy. The study has recently been amended to allow PCI to be performed before completion of all therapy. Patients are eligible if they have had a complete response, partial response, or stable disease to their treatment that must include either surgery or chemotherapy (ie, chemotherapy alone does not constitute definitive therapy). Our study suggests that the shorter the interval from the diagnosis of stage III NSCLC to the initiation of PCI, the more likely it is that PCI will reduce the incidence of brain metastases and the subsequent survival of these patients.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented at the 39th Annual Meeting of the American Society for Clinical Oncology, Chicago, IL, May 31-June 3, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Le Chavalier T, Pignon JP, Bergman B, et al: Results of the randomized International Adjuvant Lung Cancer Trial (IALT): Cisplatin-based chemotherapy (CT) vs no CT in 1867 patients (PTS) with resected non-small cell lung cancer (NSCLC). Lung Cancer 41:3, 2003 (suppl 2)
Andre F, Grunenwald D, Pujol JL, et al: Patterns of relapse of N2 nonsmall-cell lung carcinoma patients treated with preoperative chemotherapy: Should prophylactic cranial irradiation be reconsidered Cancer 91:2394每2400, 2001
Ceresoli GL, Gregorc V, Cappuzzo F, et al: ZD1839 in non-small cell lung cancer (NSCLC) patients with brain metastases (BM). Proc Am Soc Clin Oncol 22:674, 2003 (abstr 2709)
Albain KS, Rusch VW, Crowley JJ, et al: Concurrent cisplatin/etoposide plus chest radiotherapy followed by surgery for stages IIIA (N2) and IIIB non-small-cell lung cancer: Mature results of Southwest Oncology Group phase II study 8805. J Clin Oncol 13:1880每1892, 1995
Gandara DR, Chansky K, Albain KS, et al: Consolidation docetaxel after concurrent chemoradiotherapy in stage IIIB non-small-cell lung cancer: Phase II Southwest Oncology Group Study S9504. J Clin Oncol 21:2004每2010, 2003
Albain KS, Crowley JJ, Turrisi AT, et al: Concurrent cisplatin, etoposide plus radiotherapy for pathologic stage IIIB non-small cell lung cancer: A Southwest Oncology Group phase II study (S9019). J Clin Oncol 20:3454每3460, 2002
Rusch VW, Giroux DJ, Kraut MJ, et al: Induction chemoradiation and surgical resection for non-small cell lung carcinomas of the superior sulcus: Initial results of Southwest Oncology Group Trial 9416 (Intergroup Trial 0160). J Thorac Cardiovasc Surg 121:472每483, 2001
Ryan GF, Ball DL, Smith JG: Treatment of brain metastases from primary lung cancer. Int J Radiat Oncol Biol Phys 31:273每278, 1995
Sheehan JP, Sun MH, Kondziolka D, et al: Radiosurgery for non-small cell lung carcinoma metastatic to the brain: Long-term outcomes and prognostic factors influencing patient survival time and local tumor control. J Neurosurg 97:1276每1281, 2002
Robnett TJ, Machtay M, Stevenson JP, et al: Factors affecting the risk of brain metastases after definitive chemoradiation for locally advanced non每small-cell lung carcinoma. J Clin Oncol 19:1344每1349, 2001
Gaspar L, Scott C, Rotman M, et al: Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37:745每751, 1997
Agboola O, Benoit B, Cross P, et al: Prognostic factors derived from recursive partition analysis (RPA) of Radiation Therapy Oncology Group (RTOG) brain metastases trials applied to surgically resected and irradiated brain metastatic cases. Int J Radiat Oncol Biol Phys 42:155每159, 1998
Rodrigus P, de Brouwer P, Raaymakers E: Brain metastases and non-small cell lung cancer: Prognostic factors and correlation with survival after irradiation. Lung Cancer 32:129每136, 2001
Zabel A, Milker-Zabel S, Thilmann C: Treatment of brain metastases in patients with non-small cell lung cancer (NSCLC) by stereotactic linac-based radiosurgery: Prognostic factors. Lung Cancer 37:87每94, 2002
Abrahams JM, Torchia M, Putt M, et al: Risk factors affecting survival after brain metastases from non-small cell lung carcinoma: A follow-up study of 70 patients. J Neurosurg 95:595每600, 2001
D'Amico TA, Aloia TA, Moore MB, et al: Predicting the sites of metastases from lung cancer using molecular biologic markers. Ann Thorac Surg 72:1144每1148, 2001
Bubb RS, Komaki R, Hachiya T, et al: Association of Ki-67, p53, and bcl-2 expression of the primary non-small-cell lung cancer lesion with brain metastatic lesion. Int J Radiat Oncol Biol Phys 53:1216每1224, 2002(Laurie E. Gaspar, Kari Ch)
Loyola University Medical Center, Maywood IL
University of Washington Medical Center, Seattle WA
ABSTRACT
PURPOSE: A retrospective review of the Southwest Oncology Group (SWOG) database was undertaken to review the incidence and timing of diagnosis of brain metastases in patients undergoing combined-modality therapy for stage III non每small-cell lung cancer (NSCLC).
PATIENTS AND METHODS: Four hundred twenty-two eligible, assessable patients with stage IIIA/B NSCLC were treated on four SWOG protocols. Treatment varied with protocol but consisted of concurrent cisplatin-etoposide and radiation in all patients, with a surgery arm in two of the four protocols.
RESULTS: Of the 422 total patients, 268 (64%) have experienced disease progression; 54 relapses (20%) were in brain only, 17 (6.5%) were in brain and other sites simultaneously, and 197 (63.5%) were in sites other than brain. Of the 268 patients with disease progression, progression in the brain only, in the brain and other sites, and not in the brain occurred in 20%, 6%, and 74% of patients, respectively. Time from treatment to diagnosis of disease progression in the brain in 71 patients was as follows: during treatment, 16 relapses (22.5%); 0 to 16 weeks after treatment, 17 relapses (24%); 16 weeks to 6 months after treatment, 10 relapses (14%); 6 to 12 months after treatment, 16 relapses (22.5%); and more than 12 months after treatment, 12 relapses (17%). Nonsquamous histology and young patient age were the only significant predictors for increased risk of early relapse with brain metastases.
CONCLUSION: Brain metastases often develop early in the course of treatment for stage IIIA/B NSCLC. The statistical designs of ongoing trials of prophylactic cranial irradiation in stage III NSCLC have taken this into account.
INTRODUCTION
Brain metastases from non每small-cell lung cancer (NSCLC) have gained attention recently because combined-modality therapy has lead to improvements in local control and overall survival rates. With improved survival after combined-modality therapy, it has been noted that the incidence of brain metastases, often as a sole site of relapse, has increased.1每3 Andre et al,2 in a retrospective review of clinical N2 patients treated with surgery, found that the cumulative incidence of brain metastases in patients treated with or without neoadjuvant chemotherapy was 32% and 18%, respectively (P < .05). Ceresoli et al3 determined that, in stage III NSCLC patients treated with combined-modality therapy, the brain was the site of first relapse in 23% of patients and, ultimately, brain metastases developed in 50% of patients at some point in the course of their disease. This retrospective review was undertaken to determine the incidence of brain metastases, as well as predictive and prognostic factors for brain metastases in four Southwest Oncology Group (SWOG) studies of combined-modality therapy in stage III NSCLC.
PATIENTS AND METHODS
A retrospective review was undertaken of four SWOG studies in stage III NSCLC.4每7 The minimum follow-up time on these studies was 4 years. All studies required signed informed consent.
SWOG 8805 was a phase II study to assess the feasibility of and outcomes from concurrent chemotherapy and irradiation followed by surgery in pathologically proven stage IIIAN2 or IIIB NSCLC.4 The eligibility criteria included a performance status of 0, 1, or 2 and a forced expiratory volume in 1 second of 2 L or a predicted postresection forced expiratory volume in 1 second of at lease 1 L. Patients initially received two cycles of cisplatin (50 mg/m2 on days 1, 8, 29, and 36) and etoposide (50 mg/m2 on days 1 through 5 and 29 through 33) with concurrent chest radiotherapy to 45 Gy that began on day 1. Surgical resection was attempted if there was a response or stable disease. Postoperative chemoradiotherapy, consisting of two more cycles of cisplatin-etoposide with completion of radiation to total of 59.4 Gy, was administered if there was either unresectable disease or positive margins or nodes found at the time of surgery. The protocol allowed prophylactic cranial irradiation (PCI; 30 Gy in 15 fractions, 2 Gy per fraction) to be administered at the option of the investigator. To be eligible at study entry, patients had to have a negative computed tomography (CT) scan or magnetic resonance imaging (MRI) of the brain within 30 days of registration. The re-evaluation before surgery also required a CT or MRI of the brain. CT or MRI of the brain was performed at 6, 12, and 18 months, and then yearly for 3 years. The study met its accrual goal in March 1992. One hundred twenty-six patients participated in the study; 60% had stage IIIAN2 disease, and 40% had stage IIIB disease. The 18 patients who received PCI are excluded from this analysis. The median survivals for patients with stage IIIA and IIIB disease were 13 and 17 months, respectively. The 3-year survival rates for patients with stage IIIA and IIIB disease were 27% and 24%, respectively.
SWOG 9019 was initially designed as a phase III study in patients with pathologic stages IIIAN2 and IIIB NSCLC. Eligibility criteria were identical to SWOG 8805, as discussed earlier. Eligible patients had to have a normal CT or MRI of the brain within 30 days of study registration. Patients were randomly assigned to either arm 1 (two cycles of cisplatin 50 mg/m2 on days 1, 8, 29, and 36 and etoposide 50 mg/m2 on days 1 through 5 and 29 through 33 with concurrent thoracic radiation to 61 Gy) or arm 2, which was the identical trimodality therapy as administered in SWOG 8805. All patients received two more cycles of cisplatin and etoposide using the same dose and schedule as used during the concurrent portion of treatment. CT or MRI of the brain was required before consolidation chemotherapy only if new symptoms developed. A CT or MRI of the brain was required at 12, 18, and 24 months of follow-up, and then yearly thereafter. Shortly into its accrual, the enrollment of all patients with stage IIIAN2 disease onto both arms of patients with stage IIIB disease onto the surgical arm was discontinued to support one North American trial addressing the surgical question (INT 0139). Thus, SWOG 9019 became a single-arm study of definitive chemoradiotherapy for patients with stage IIIB disease, and accrual was completed and reported as such.6 There were 50 eligible patients with the following tumor substages: 18 patients, T4N0/1; 12 patients, T4N2; and 20 patients, N3. The overall median survival time was 15 months, and the 3-year survival rate was 17%. For this report on brain metastases, however, all of the originally enrolled patients (both arms, both stages) are included because mature follow-up and site of relapse data are available.
SWOG 9416 was a phase II study in patients with NSCLC of the superior sulcus.7 This trial tested the feasibility of induction chemoradiotherapy and surgical resection in NSCLC of the superior sulcus. Eligible patients required a normal CT or MRI of the brain within 42 days of registration. Patients with T3-4N0-1 superior sulcus NSCLC received two cycles of cisplatin (50 mg/m2 on days 1, 8, 29, and 36) with etoposide (50 mg/m2 on days 1 through 5 and 29 through 33). This chemotherapy was concurrent with 45 Gy of thoracic radiation therapy. Patients with stable or responding disease underwent thoracotomy 3 to 5 weeks later. A CT or MRI of the brain was required before thoracotomy. Two more cycles of cisplatin-etoposide were to be administered postoperatively. The study was closed in August 1999. The 2-year survival rate was 55% for all eligible patients and 70% for patients who underwent a complete resection.
SWOG 9504 was a phase II study in patients with pathologically proven stage IIIB NSCLC that was designed to test the concept of taxane sequencing in combined-modality therapy. The eligibility criteria were identical to SWOG 9019. Eligible patients had to have a CT or MRI of the brain documenting no brain metastases within 42 days of registration. Patients received cisplatin (50 mg/m2 on days 1, 8, 29, and 36) and etoposide (50 mg/m2 on days 1 through 5 and 29 through 33). This chemotherapy was administered with concurrent thoracic radiation to a dose of 61 Gy, given in 1.8-Gy daily fractions to 45 Gy and then 2.0-Gy daily fractions for an additional 16 Gy. Consolidation docetaxel started 4 to 6 weeks after completion of concurrent chemoradiotherapy. A CT or MRI of the brain was required before consolidation chemotherapy only if there were new symptoms. The docetaxel dose started at 75 mg/m2 but was to be increased to 100 mg/m2 in cycles 2 and 3 if tolerated.5 During follow-up, a CT or MRI of the brain was required at 12, 18, and 24 months, and then yearly thereafter. SWOG 9504 closed in December 1998 after having accrued 83 eligible patients with the following stage subsets: T4N0-1, 31 patients (37%); T4N2, 22 patients (27%), and T1-3N3, 30 patients (36%). The median progression-free survival time was 16 months. The survival rates at 1, 2, and 3 years were 76%, 54%, and 37%, respectively.
A statistical analysis was performed to look at the incidence and timing of development of brain metastases in the 422 eligible, assessable patients enrolled onto the four protocols, with the exception of the 18 patients on SWOG 8805 who were known to have received PCI. The analysis includes patients for whom progression was never observed or recorded. For those patients who progressed but for whom the site of progression was not recorded, often because the patients died of their disease before restaging, it is assumed that progression was at a site other than the brain. Logistic regression was performed with the outcome being brain metastases (yes or no). A Cox regression analysis, with the outcome being time to brain metastases (with censoring for all other outcomes including progression at a site other than brain), was also performed. Failure during treatment includes failures that occurred during any phase of treatment (ie, concurrent chemoradiotherapy or surgery or consolidation chemotherapy). Only the site(s) of first progression was recorded. The following variables were entered into univariate analyses as well as a proportional hazards model with stepwise selection: sex, stage (IIIA v IIIB), performance status (Eastern Cooperative Oncology Group performance status of 0 v 1 v 2), histology, planned docetaxel consolidation, age, and planned surgery (yes or no). Estimates for the risk of brain relapse were generated using cumulative incidence with competing risks for death or progression to other sites.
RESULTS
There were 422 eligible patients on the four SWOG studies, of whom 64% have had documented tumor progression or relapse. Table 1 lists the total number of patients per study and the percentage of patients in each study in whom progression was recorded. Progression was categorized as brain only (brain metastases with no other site of failure), brain and other sites (brain metastases with simultaneous failure in extracranial site), or not brain (extracranial failure only). Progression of the primary lung cancer was counted as an extracranial failure. Progression expressed as a percentage of the 422 total patients was found to be brain only, brain and other sites, and not brain in 13%, 4%, and 47% of patients, respectively (Table 2). Of the 268 patients with progression, progression in the brain only, in the brain and other sites, and not in the brain occurred in 20%, 6%, and 74% of patients, respectively (Table 3).
We examined potential predictors of progression site using both univariate and stepwise logistic regression using the following variables: SWOG study, planned surgery (yes or no), planned docetaxel consolidation (yes or no), stage (IIIB v IV), age, and histology. The only variables that reached statistical significance were histology and patient age at presentation. The incidence of disease progression according to histology is summarized in Table 4. There was no significant risk of overall progression according to histology. However, nonsquamous histology was associated with a higher risk of progression within the brain, either as the sole site of relapse or concurrent with failure at other sites (P = .007, odds ratio = 2.4 for nonsquamous v squamous histology; Table 5).
Table 6 summarizes the time of first failure relative to study registration for the 268 patients who have had documented progression. This table also indicates the time of failure according to the site of progression. Seventy-eight patients (29.1% of all patients who experienced treatment failures) had documented progression during the course of their therapy. Eighty percent of these failures were at distant sites excluding the brain. Time from study registration to the diagnosis of brain metastases in the 71 patients who developed brain metastases was as follows: during treatment, 16 patients (22.5%); 0 to 16 weeks after treatment, 17 patients (24%); 16 weeks to 6 months after treatment, 10 patients (14%); 6 to 12 months after treatment, 16 patients (22.5%); and more than 12 months after treatment, 12 patients (17%). The time of first failure was also calculated from the date of study registration. Time from study registration to the diagnosis of brain metastases in the 71 patients who developed brain metastases was as follows: 0 to 6 months from registration, 26 patients (37%); 6 to 12 months from registration, 23 patients (32%); and more than 12 months from registration, 22 patients (31%).
Brain only as a site of failure was found in 54 patients (20% of all patients who experienced treatment failures), and 14% of the failures occurred during treatment. Almost half of all failures (46.5%) within the brain (brain only or brain and other sites) occurred during or within 16 weeks of completing therapy. Nonsquamous histology was associated with a shorter time to relapse in the brain compared with squamous carcinomas. Multivariable Cox regression analysis yielded results similar to the logistic regression analysis. Histology and patient age were the only significant predictors for brain metastases among the variables examined. The hazard ratio was 2.1 for nonsquamous histology versus squamous histology (P = .02). Younger patients were at increased risk for developing brain metastases. The hazard ratio was 1.8 for age 50 years versus more than 50 years (P = .046), Factors that were not found to predict time of failure were intent for surgery, sex, use of taxane consolidation, performance status, and stage (IIIA v IIIB).
The median survival times measured from the date of study registration were 26, 12, and 16 months for patients with relapse in brain only, relapse in brain and other sites, and distant failure not in brain, respectively (Fig 1). The median survival times measured from the date of progression were 10.3, 4.5, and 5 months for patients with relapse in brain only, relapse in brain and other sites, and distant failure not in brain, respectively. The median survival after progression of 10.3 months for patients with relapse in the brain only was significantly different from the 5-month survival of patients with all other sites of failure (P = .0002). The interval from study registration to tumor progression was a factor predictive of subsequent survival. Median survival time after progression for patients who experienced treatment failure within 6 months or after more than 6 months from study registration was 5 and 8 months, respectively (P = .05).
The overall 2-year cumulative incidence of brain failure was 17% (Fig 2). Patients with adenocarcinoma had the highest rate of brain failure, with a cumulative incidence of 22% v only 10% for patients with squamous cell histology (Fig 3). Patients in the 50- to 60-year age range had a comparable risk of brain failure (12% to 16% 2-year incidence), whereas patients presenting at age less than 50 years were much more likely to have the brain as their first site of failure, with a 2-year incidence of 26%.
DISCUSSION
This analysis found that, for stage III NSCLC, the factors on multivariate analysis that predicted for the subsequent development of brain metastases were histology and age less than 50 years. Patients with nonsquamous histology were more likely than patients with squamous cell histology to progress at any site and also had an increased risk of having relapse with brain metastases. The 2-year cumulative risk of failure in the brain for patients with adenocarcinoma and squamous cell carcinoma was 22% and 10%, respectively. The influence of histology on the risk of brain metastases has been observed by others.2,8,9 Ryan et al8 reported the breakdown by histology in one large series of patients with brain metastases from lung cancer as follows: 31% small-cell carcinoma, 21% adenocarcinoma, 21% large-cell carcinoma, and 8% squamous cell carcinoma. Andre et al2 reported the incidence of isolated brain metastases in patients treated with cisplatin-based neoadjuvant chemotherapy and surgery. Isolated brain metastases occurred in 6%, 28%, and 8% of patients with squamous cell carcinoma, adenocarcinoma, and other histologies, respectively (P .05). This study found the following factors not to be predictive of subsequent development of brain metastases: sex, performance status, and stage (IIIA v IIIB).
Other studies have found a range of factors to be significant. Ceresoli et al3 reported that stage III NSCLC patients were more likely to develop brain metastases if they were younger than 60 years of age. This observation is in agreement with our observation of an increased risk of brain metastases in patients less than age 50 years of age. Ceresoli et al3 also found that the bulk of mediastinal lymphadenopathy (> 2 cm v 2 cm) was a significant factor on univariate analysis but that it lost its significance on multivariate analysis. Our study was unable to address this issue of bulk of mediastinal lymphadenopathy.
The choice of treatment likely affects the pattern of tumor relapse. Andre et al2 conducted a retrospective review and concluded that the risk of brain metastases was increased in patients who received neoadjuvant chemotherapy followed by surgery. The 3-year incidence of isolated brain metastases in patients with clinical N2 NSCLC treated with or without neoadjuvant chemotherapy was 20% and 9%, respectively. Robnett et al10 found that the sequencing of chemotherapy and radiation therapy in locally advanced NSCLC was associated with significantly different rates of subsequent brain metastases. The crude and 2-year actuarial risk of brain metastases was 27% and 39%, respectively, when induction chemotherapy was followed by radiation therapy. For patients having immediate concurrent chemoradiotherapy, the crude and 2-year actuarial risk of brain metastases was 15% and 20%, respectively. Our study found that the use of taxane consolidation did not affect the subsequent risk of developing brain metastases.
Our study determined that the survival rate after diagnosis of brain metastases was significantly longer for patients with failure in the brain only as opposed to failure simultaneously in the brain and other sites. This is in keeping with other retrospective reviews, including the recursive partitioning analysis (RPA) by the Radiation Therapy Oncology Group (RTOG).11每14 The RTOG RPA examined prognostic factors in 1,200 patients with brain metastases from all primary solid tumors treated on RTOG studies.11 Sixty-one percent of those patients had primary lung cancer, with only a small number of patients with small-cell histology. Variables found to be of significant prognostic importance were performance status, extent of extracranial disease, control of primary cancer, and age. In another relatively large retrospective review of 250 patients with brain metastases from NSCLC, patients with an absent or controlled primary tumor had a 1-year survival rate of 26% as opposed to just 11% for patients with an active primary tumor (P = .051).13
This study found that patients who developed brain metastases within 6 months of study registration did significantly worse than patients who developed brain metastases at a later date (median survival, 8 v 6 months; P = .05). The prognostic impact of the time interval between the diagnosis of the primary NSCLC and brain metastases has been found in other retrospective series.9,15 Abrahams et al,15 at the Hospital of the University of Pennsylvania, reported an improved outcome for patients with brain metastases diagnosed more than 3 months from diagnosis of the primary cancer compared with patients with brain metastases diagnosed earlier. The median survival time and 5-year survival rate for metachronous versus synchronous brain metastases were 18 months and 28.9% v 9.9 months and 0%, respectively. The interval between diagnosis and developing brain metastases was also a factor predictive of survival in a series of NSCLC patients treated with radiosurgery.9 The median survival time from the date of radiosurgery was 11 months for patients diagnosed more than 7 months from diagnosis of primary cancer compared with just 7 months for patients with a shorter interval. The RTOG RPA found that the interval from diagnosis of the primary cancer to the development of brain metastases ( 2 years v > 2 years) was of no prognostic importance.11 This discrepancy between the RTOG RPA conclusions and those of other studies could have been a result of the time point chosen (ie, 2 years in the RTOG analysis v 3 to 7 months in other studies).
The most important observation from this study is that the interval between study registration and the development of brain metastases was so short. Forty-six percent of patients who developed brain metastases had these diagnosed within 16 weeks of completing treatment. The two factors on multivariate analysis that were predictive of a short interval to the development of brain metastases were nonsquamous histology and young patient age. The hazard rate for brain metastases was significantly higher in patients with adenocarcinoma or other nonsquamous histologies and significantly higher for patients who were age 50 years or younger.
There have been few reports that deal with the issue of timing of the development of brain metastases in patients with initially locally advanced NSCLC. Ceresoli et al3 reported the time to brain relapse measured from the initiation of treatment in 112 patients with stage IIB, IIIA, or IIIB NSCLC. All patients were treated with neoadjuvant chemotherapy (cisplatin, vinblastine, and mitomycin) and then re-evaluated for surgery. Patients considered to have potentially resectable disease went on to surgery and postoperative radiation. Unresectable patients were treated with radiation alone. The median time to brain relapse measured from the initiation of therapy was just 9 months. The only two factors predictive of a short time to relapse within the brain in multivariate analysis were young age (less than 60 years) and mediastinal lymphadenopathy greater than 2 cm in diameter. The median time to brain relapse in patients with squamous and nonsquamous histology was 21 and 14 months, respectively, although this difference did not reach statistical significance.
Using standard patient and tumor characteristics as predictors of relapse within the brain has obvious limitations in terms of positive predictive value. Another approach is to look at molecular markers. Investigators have looked at molecular markers and their association with the subsequent occurrence of brain metastases in NSCLC.16,17 D'Amico et al16 retrospectively reviewed 202 pathologic specimens from patients undergoing resection of stage I NSCLC, of whom 25 were subsequently found to have isolated brain metastases. A panel of eight molecular markers was chosen to represent multiple pathways in the metastatic process: growth regulation (erbB2), apoptosis (p53), angiogenesis (factor viii), cellular adhesion (EphA2 and E-cadherin), and basement membrane invasion (UPA, UPAR, and PAI-1). Patients with isolated brain relapse had significantly higher expression of p53 and UPA. In addition, patients with E-cadherin expression had a low incidence of developing brain metastases. Bubb et al17 found that Ki-67, p53, and bcl-2 in primary NSCLC primary tumors were not predictive of survival or the incidence of brain metastases. There needs to be more work done in this area.
In conclusion, this study demonstrated a high rate of early relapse within the brain in patients with stage III NSCLC treated with multimodality therapy. Many of the failures were isolated brain metastases. These results argue for some form of treatment to prevent the development of brain metastases. PCI has been proposed as one such method of reducing the incidence of brain metastases. Many cooperative groups within North America have joined the RTOG-led phase III study in which patients with stage III NSCLC are randomly assigned to either receive or not receive PCI. RTOG L-0214 was originally open to patients with stage IIIA or IIIB NSCLC who had completed all planned definitive locoregional therapy. Patients could be randomized to PCI or no PCI for up to 16 weeks after completion of all therapy. The study has recently been amended to allow PCI to be performed before completion of all therapy. Patients are eligible if they have had a complete response, partial response, or stable disease to their treatment that must include either surgery or chemotherapy (ie, chemotherapy alone does not constitute definitive therapy). Our study suggests that the shorter the interval from the diagnosis of stage III NSCLC to the initiation of PCI, the more likely it is that PCI will reduce the incidence of brain metastases and the subsequent survival of these patients.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented at the 39th Annual Meeting of the American Society for Clinical Oncology, Chicago, IL, May 31-June 3, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Le Chavalier T, Pignon JP, Bergman B, et al: Results of the randomized International Adjuvant Lung Cancer Trial (IALT): Cisplatin-based chemotherapy (CT) vs no CT in 1867 patients (PTS) with resected non-small cell lung cancer (NSCLC). Lung Cancer 41:3, 2003 (suppl 2)
Andre F, Grunenwald D, Pujol JL, et al: Patterns of relapse of N2 nonsmall-cell lung carcinoma patients treated with preoperative chemotherapy: Should prophylactic cranial irradiation be reconsidered Cancer 91:2394每2400, 2001
Ceresoli GL, Gregorc V, Cappuzzo F, et al: ZD1839 in non-small cell lung cancer (NSCLC) patients with brain metastases (BM). Proc Am Soc Clin Oncol 22:674, 2003 (abstr 2709)
Albain KS, Rusch VW, Crowley JJ, et al: Concurrent cisplatin/etoposide plus chest radiotherapy followed by surgery for stages IIIA (N2) and IIIB non-small-cell lung cancer: Mature results of Southwest Oncology Group phase II study 8805. J Clin Oncol 13:1880每1892, 1995
Gandara DR, Chansky K, Albain KS, et al: Consolidation docetaxel after concurrent chemoradiotherapy in stage IIIB non-small-cell lung cancer: Phase II Southwest Oncology Group Study S9504. J Clin Oncol 21:2004每2010, 2003
Albain KS, Crowley JJ, Turrisi AT, et al: Concurrent cisplatin, etoposide plus radiotherapy for pathologic stage IIIB non-small cell lung cancer: A Southwest Oncology Group phase II study (S9019). J Clin Oncol 20:3454每3460, 2002
Rusch VW, Giroux DJ, Kraut MJ, et al: Induction chemoradiation and surgical resection for non-small cell lung carcinomas of the superior sulcus: Initial results of Southwest Oncology Group Trial 9416 (Intergroup Trial 0160). J Thorac Cardiovasc Surg 121:472每483, 2001
Ryan GF, Ball DL, Smith JG: Treatment of brain metastases from primary lung cancer. Int J Radiat Oncol Biol Phys 31:273每278, 1995
Sheehan JP, Sun MH, Kondziolka D, et al: Radiosurgery for non-small cell lung carcinoma metastatic to the brain: Long-term outcomes and prognostic factors influencing patient survival time and local tumor control. J Neurosurg 97:1276每1281, 2002
Robnett TJ, Machtay M, Stevenson JP, et al: Factors affecting the risk of brain metastases after definitive chemoradiation for locally advanced non每small-cell lung carcinoma. J Clin Oncol 19:1344每1349, 2001
Gaspar L, Scott C, Rotman M, et al: Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37:745每751, 1997
Agboola O, Benoit B, Cross P, et al: Prognostic factors derived from recursive partition analysis (RPA) of Radiation Therapy Oncology Group (RTOG) brain metastases trials applied to surgically resected and irradiated brain metastatic cases. Int J Radiat Oncol Biol Phys 42:155每159, 1998
Rodrigus P, de Brouwer P, Raaymakers E: Brain metastases and non-small cell lung cancer: Prognostic factors and correlation with survival after irradiation. Lung Cancer 32:129每136, 2001
Zabel A, Milker-Zabel S, Thilmann C: Treatment of brain metastases in patients with non-small cell lung cancer (NSCLC) by stereotactic linac-based radiosurgery: Prognostic factors. Lung Cancer 37:87每94, 2002
Abrahams JM, Torchia M, Putt M, et al: Risk factors affecting survival after brain metastases from non-small cell lung carcinoma: A follow-up study of 70 patients. J Neurosurg 95:595每600, 2001
D'Amico TA, Aloia TA, Moore MB, et al: Predicting the sites of metastases from lung cancer using molecular biologic markers. Ann Thorac Surg 72:1144每1148, 2001
Bubb RS, Komaki R, Hachiya T, et al: Association of Ki-67, p53, and bcl-2 expression of the primary non-small-cell lung cancer lesion with brain metastatic lesion. Int J Radiat Oncol Biol Phys 53:1216每1224, 2002(Laurie E. Gaspar, Kari Ch)