J Clin Oncol 22: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 22 NUMBER 1 JANUARY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T From the M.D. Anderson Cancer Center, Houston, TX; Pharmacyclics Inc, Sunnyvale; California Cancer Care, Greenbrae, CA; University of Wisconsin Medical School, Madison, WI; New Mexico Hematology-Oncology Consultants, Albuquerque, NM; North Memorial Research Center, Robbinsdale, MN; University of Colorado, Denver, CO; Thomas Jefferson University, Philadelphia, PA; Princess Margaret Hospital, Toronto; Ottawa Regional Cancer Centre, Ottawa, Canada; Wessex Cancer Centre, Southampton, United Kingdom; Western General Hospital, Edinburgh, Scotland; Medisch Spectrum Twente, Enschede, The Netherlands; and Centre Léon Bérard, Lyon, France. Submitted May 19, 2003; accepted October 30, The study drug and funding for this research were provided by Pharmacyclics Inc. This study was presented in part at the 38 th Annual Meeting of the American Society of Clinical Oncology, May 18-21, 2002, Orlando, FL; 44 th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, October 6-10, 2002, New Orleans, LA; and 7 th Annual Meeting of the Society of Neuro-Oncology, November 21-24, 2002, San Diego, CA. C.A.M and J.A.S. contributed equally to this manuscript. Neurocognitive Function and Progression in Patients With Brain Metastases Treated With Whole-Brain Radiation and Motexafin Gadolinium: Results of a Randomized Phase III Trial Christina A. Meyers, Jennifer A. Smith, Andrea Bezjak, Minesh P. Mehta, James Liebmann, Tim Illidge, Ian Kunkler, Jean-Michel Caudrelier, Peter D. Eisenberg, Jacobus Meerwaldt, Ross Siemers, Christian Carrie, Laurie E. Gaspar, Walter Curran, See-Chun Phan, Richard A. Miller, and Markus F. Renschler A B S T R A C T Purpose To report the neurocognitive findings in a phase III randomized trial evaluating survival and neurologic and neurocognitive function in patients with brain metastases from solid tumors receiving whole-brain radiation therapy (WBRT) with or without motexafin gadolinium (MGd). Patients and Methods Patients were randomly assigned to receive WBRT 30 Gy in 10 fractions with or without MGd 5 mg/kg/d. Monthly neurocognitive testing for memory, executive function, and fine motor skill was performed. Results Four hundred one patients were enrolled (251 with non small-cell lung cancer, 75 with breast cancer, and 75 with other cancers); 90.5% patients had impairment of one or more neurocognitive tests at baseline. Neurocognitive test scores of memory, fine motor speed, executive function, and global neurocognitive impairment at baseline were correlated with brain tumor volume and predictive of survival. There was no statistically significant difference between treatment arms in time to neurocognitive progression. Patients with lung cancer (but not other types of cancer) who were treated with MGd tended to have improved memory and executive function (P.062) and improved neurologic function as assessed by a blinded events review committee (P.048). Conclusion Neurocognitive tests are a relatively sensitive measure of brain functioning; a combination of tumor prognostic variables and brain function assessments seems to predict survival better than tumor variables alone. Although the addition of MGd to WBRT did not produce a significant overall improvement between treatment arms, MGd may improve memory and executive function and prolong time to neurocognitive and neurologic progression in patients with brain metastases from lung cancer. J Clin Oncol 22: by American Society of Clinical Oncology Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Christina A. Meyers, PhD, M.D. Anderson Cancer Center, Department of Neuro-Oncology, Unit 431, 1515 Holcombe Blvd, Houston, TX 77030; cameyers@mdanderson.org by American Society of Clinical Oncology X/04/ /$20.00 DOI: /JCO INTRODUCTION Brain metastases are a frequent complication of many cancers, occurring in as many as 24% of all patients on the basis of autopsy data [1]. In lung cancer, the most common cause of brain metastases, up to 50% of patients develop CNS involvement [2-4]. Brain metastases occur early in lung cancer and sometime produce neurologic symptoms at disease presentation, in contrast to other cancers in which CNS spread is usually a later complication [1]. Other tumor types commonly associated with brain metastases include breast cancer (15% to 20%), unknown primary tumor (2% to 15%), melanoma (10%), and colon cancer (5%) [5-9]. The median survival of patients with brain metastases treated with whole-brain radiation therapy (WBRT) is only approxi- 157

2 Meyers et al mately 4 months [10-12]; clinical trials comparing various radiation therapy regimens and radiosensitizers have not demonstrated a survival benefit [13-19]. The majority of patients with brain metastases have neurologic and neurocognitive impairment [20]. In patients with primary brain tumors, neurocognitive impairments preventing functional independence have been shown to be more common than physical disability [21]. Any treatment that would reduce the severity of neurologic and neurocognitive impairments would therefore potentially enhance the quality of life of patients with brain metastases. Brain metastases not amenable to surgical resection are often treated with radiation therapy. However, normal tissue tolerance limits radiation dosage. Use of a tumor-selective agent that enhances radiation effect in tumors but spares normal brain might improve the therapeutic ratio of WBRT, improving local control of the tumor without increasing radiation toxicity. Motexafin gadolinium (MGd) targets tumors selectively and generates reactive oxygen species intracellularly, lowering the apoptotic threshold to radiation and chemotherapy. It increases tumor radiation response in vivo in preclinical models [22-24]. MGd is paramagnetic, and previous clinical studies have demonstrated tumor localization using magnetic resonance imaging (MRI) [25-27]. A randomized phase III study was conducted to determine if MGd, administered with WBRT, would improve outcome for patients with brain metastases [28]. Because patients with brain metastases frequently die as a result of systemic disease progression, a survival end point may be of limited value in assessing the clinical benefit of a new treatment for brain metastases. This study, therefore, employed tests of neurologic function and neurocognitive function (NCF) to evaluate the effect of improved local tumor control. Although, as reported elsewhere [28], no significant difference in survival was observed between the two treatment arms, differences in time to neurologic progression and NCF were observed between the two treatment arms. The relationship between NCF and clinical variables, as well as the results of treatment with MGd on NCF, are summarized in this report. This is the first large clinical trial in which neurocognitive assessments have been studied in patients with brain metastases. PATIENTS AND METHODS Study Design The study design, treatment, and assessments have been described previously [28]. In brief, this was a prospective, open-label, multicenter clinical trial (institutions and investigators are listed in the Appendix). The protocol was reviewed and approved by institutional ethics committees at each participating center in accordance with the Helsinki Declaration. Written informed consent was obtained from each patient before participation in the trial. Adult patients were eligible to participate if they had an MRI that demonstrated brain metastases from histologically proven solid tumors, required WBRT, and had a Karnofsky performance score (KPS) 70. Patients were excluded if they had small-cell lung cancer, lymphoma, or germ cell tumors; had brain metastases that had been resected; had received prior cranial irradiation, including stereotactic radiosurgery; had leptomeningeal or liver metastases; or had two or more sites of extracranial metastases, except when breast was the primary cancer. Treatments Patients were randomly assigned to receive 30 Gy WBRT given in 10 daily fractions, with or without MGd 5 mg/kg/d, 2 to 5 hours before each fraction of WBRT, as has been previously published [28]. Guidelines for the initiation and tapering of corticosteroids were specified by the protocol. Efficacy Assessments The coprimary end points of the study were survival and time to neurologic progression. Time to neurocognitive progression was measured as a secondary end point in this trial. Patients were evaluated at entry, at monthly intervals for the first 6 months, and then every 3 months until death. MRI scans were obtained at baseline, at 2, 4, and 6 months, and then every 3 months, and were reviewed in a blinded manner [28]. Up to six measurable brain metastases were selected at presentation as indicator lesions, the sum of which is referred to in the text and tables that follow as indicator lesion volume. Data on postrandomization use of corticosteroids, anticonvulsants, and anticancer therapies were collected. All patients underwent a battery of standardized neurocognitive tests administered by trained and certified nurses or clinical research associates at each study visit [28]. All individuals administering these tests underwent thorough training by review of training manuals, videotapes, and hands-on instruction followed by certification. Completed neurocognitive tests were scored centrally by a blinded, central reviewer to remove potential treatment bias in assessing outcomes. z scores were derived from the patient s scores in the individual neurocognitive tests, using an age-adjusted (and for controlled oral word association [COWA], education-adjusted) normative distribution of scores from an unimpaired population. When a patient could not attempt or did not complete any of these tests, the primary cause of the missed assessment was recorded. Causes considered by investigators to be related to a disability resulting from a brain metastasis were included as progression in the analysis. Patients were considered to be impaired in a particular neurocognitive test at baseline if their z score was 1.5 standard deviations (SDs) worse than the mean of that test s normative age-adjusted distribution. Statistical Methods and Sample Size Calculation All analyses were carried out on an intent-to-treat basis. No patients were excluded from the analysis. Patients were stratified at random assignment by primary tumor (breast v lung v other cancers) and the Radiation Therapy Oncology Group s recursive partitioning analysis (RPA) class (RPA class 1 v 2), determined by age, extent of extracranial disease, and status of the primary tumor [12]. The study was not powered to detect a difference in NCF, which was a secondary end point. Patients were observed for 6 months after the 400th patient was enrolled or until death. 158 JOURNAL OF CLINICAL ONCOLOGY

3 Motexafin Gadolinium in Brain Metastases Predictors of neurocognitive impairment at baseline. Univariate and stepwise logistic regression analyses were conducted to identify potential predictors of neurocognitive impairment (binary variable; three or three tests impaired) at baseline. In addition, Spearman correlations were determined to assess potential correlations between neurocognitive impairments at baseline and the indicator lesions tumor volumes and the number of brain metastases at baseline. Similarly, Spearman correlations were determined to assess potential correlations between changes in NCF from baseline to month 2 and changes in the indicator lesions tumor volume. Neurocognitive progression analysis. Tests were grouped into three domains of functioning: three memory domain tests were the Hopkins Verbal Learning Test (HVLT) for immediate recall, delayed recall, and recognition [29]; two executive function tests were COWA [30] and Trailmaking Test B [31]; two fine motor tests were pegboard dominant hand, pegboard nondominant hand [31]. Visual-motor scanning speed was measured by Trailmaking Test A [31]. Time-to-progression analysis used three progression definitions for each of the eight tests, thereby increasing the power and generalizability of the results. The three progression definitions were a worsening in each test s score from baseline by 3, 4.5, or 6 SDs in the test s normative distribution. In addition, each worsening only counted as a progression if it was confirmed by a subsequent assessment or if it was the last measurement before termination. Time to neurocognitive progression was analyzed using SUDAAN software (Release 8.0.1, Research Triangle Institute, Research Triangle Park, NC) by a repeated measures time-toevent Cox regression analysis with the tests in each domain and with the three levels of progression included as repeated measures. Patients were censored at the earliest occurrence of termination from the study, additional brain-related treatment, or death. Survival analysis. Factors predictive of survival were analyzed using Cox regression analysis. Both univariate and stepwise multivariate analyses were conducted. The following variables were included in the models: site of the primary tumor (lung v breast); sex; baseline albumin (low v normal); low, normal, or high baseline serum lactate dehydrogenase (LDH); KPS (70 to 80 v 90 to 100); age; baseline hemoglobin (low v normal); volume of indicator brain metastases; number of assessable brain metastases ( two v two); country; time from primary tumor diagnosis to enrollment; presentation with brain metastases at the time of primary cancer diagnosis; tumor histology; presence of extracranial metastasis (yes or no); number of prior chemotherapy regimens; primary tumor status (controlled or uncontrolled); number of prior chemotherapy cycles; RPA (class 1 v 2); neurocognitive tests for memory (HVLT recognition, recall, delayed recall); motor speed and dexterity (grooved pegboard test, dominant and nondominant hand); executive function (Trailmaking Test B); verbal fluency and executive function (COWA), and global neurocognitive impairment ( three tests impaired). RESULTS Patient Characteristics A total of 401 patients (251 patients with lung cancer, 75 patients with breast cancer, and 75 patients with other cancers) were randomly assigned to receive either MGd and WBRT (193 patients) or WBRT alone (208 patients). The two groups were balanced with respect to known prognostic features (Table 1). Brain was the only site of metastases in 50.8% of patients and the majority (80.1%) had multiple metastases. The majority of patients enrolled onto the study had non small-cell lung cancer (NSCLC; n 251). As previously reported, the clinical features of patients with lung cancer were found to differ significantly from patients with breast or other cancers [28]. More lung cancer patients had brain metastases at the time that their primary tumors were diagnosed, had brain as the only site of metastases, had a much shorter median time from primary cancer diagnosis to enrollment, had less prior systemic therapy, and had smaller median indicator lesion volume in the brain. There was no difference in the postrandomization use of corticosteroids, anticonvulsants, or anticancer therapy by treatment arm. Compliance Baseline neurocognitive tests were completed by 87% to 98% of patients (Table 2). Completion rates of neurocognitive tests remained high at subsequent visits. For example, 77% to 87% of patients completed the tests at their 6-month visit. NCF Test Results Baseline NCF was balanced between treatment arms for the overall study population and for each tumor strata. NCF was commonly impaired in patients at baseline; in particular, fine motor control, executive function, and memory (Table 2). Nearly all patients had some evidence of neurocognitive impairment at baseline: 363 (90.5%) patients had impairment of one or more neurocognitive tests at baseline, with 170 (42.4%) of the patients having impairment in four or more tests. Analyses of NCF versus the volume of indicator lesions at baseline and the number of brain metastases at baseline are shown in Table 3. NCF at baseline was not statistically correlated with the number of brain metastases at baseline, but each of seven neurocognitive tests demonstrated a high degree of correlation with the volume of the indicator lesion at baseline (P.0036 for memory recognition, P.0002 for COWA, and P.0001 for each of the other five neurocognitive tests; Table 3). Univariate and stepwise logistic regression analyses of global neurocognitive impairment (binary variable, three or three tests impaired) at baseline were performed using the following variables: age, sex, smoking history, prior chemotherapy (yes or no), number of prior chemotherapy cycles, number of prior chemotherapy regimens, prior neurotoxic chemotherapy (eg, vincristine or cisplatin; yes or no), number of brain metastases, time from primary tumor diagnosis to enrollment, and baseline indicator lesion volume. Lesion volume was the only predictor of global neurocognitive impairment at baseline 159

4 Meyers et al Characteristic Table 1. Patient Characteristics WBRT (n 208) No. of Patients % MGd and WBRT (n 193) No. of Patients % Total (N 401) No. of Patients % Age, years Median th percentile th percentile Sex Male Female Primary tumor Lung Breast Other RPA class Primary controlled Yes No Number of brain lesions Sites of extracranial metastases None, brain only other metastatic site or more metastatic sites KPS Serum LDH, /L Mean units SD Time from primary cancer diagnosis to enrollment, months Median th percentile th percentile No. of prior chemotherapy cycles Mean SD Abbreviations: WBRT, whole-brain radiation therapy; MGd, motexafin gadolinium; RPA, recursive partitioning analysis; KPS, Karnofsky performance score; LDH, lactate dehydrogenase; SD, standard deviation. in this model (P.0001 for the univariate analysis, and P.0014 for stepwise analysis). Time to Neurocognitive Progression Analysis of time to neurocognitive progression by treatment arm did not reach statistical significance for the overall study population. However, in the prespecified stratum of patients with brain metastases as a result of NSCLC, patients who were treated with MGd and WBRT trended to have a prolonged time-to-neurocognitive progression for memory and executive function compared with patients treated with WBRT alone (P.062 for the combined analysis of memory and executive function using six tests). Hazard ratios (HRs) for each domain and combined domains indicated a benefit in favor of the MGd treatment arm; HRs for the memory and executive function tests (separately and combined) 160 JOURNAL OF CLINICAL ONCOLOGY

5 Motexafin Gadolinium in Brain Metastases Recall Table 2. Baseline Neurocognitive Test Scores (N 401) Recognition Delayed Recall Verbal Fluency: COWA Pegboard: Dominant Hand Pegboard: Nondominant Hand Executive Function: Trail B No. of patients % of patients % of patients impaired Abbreviation: COWA, controlled oral word association. z score 1.5 below normative mean. ranged from 0.52 to 0.59, and the HR for all eight neurocognitive tests together was 0.66 (Table 4). Results for time to progression in recall memory and in COWA tests are shown for the lung cancer patients by treatment arms in Figures 1 and 2, respectively. No benefit was found in patients with breast or other cancers. Neurocognitive test results deteriorated over time. Patients progressed ( 2 SD change) most in pegboard performance (31% at 3 months) and least in COWA (7% at 3 months). Neurocognitive Test Scores and Lesion Volume Changes in NCF 2 months after treatment correlated with indicator lesion volume (Fig 3); patients with progressive disease had greater median changes in z scores for each neurocognitive test than did patients with partial response. Only patients with partial response demonstrated improvement in executive function (Trail B) and visual motor scanning (Trail A) tests. NCF and Survival Analysis of neurocognitive test scores at baseline as univariate predictors of survival is shown in Table 5. Neurocognitive tests at baseline that were predictive of survival in univariate analysis included memory (HVLT recognition, recall, and delayed recall), motor speed and dexterity (pegboard test for dominant and nondominant hands), executive function (Trailmaking Test B), and global neurocognitive impairment ( three tests impaired; Table 5). A multivariate Cox proportional hazards analysis of survival with all eight neurocognitive tests included in the model demonstrated that HVLT recognition and recall remained as independent predictors of survival (P.0323 and P.0342, respectively). When clinical as well as neurocognitive parameters are included, the pegboard dominant hand test at baseline was predictive of survival in a stepwise multivariate analysis (P.0233; Table 6). Other factors predictive of survival included sex, number of brain metastases (one to two v two), KPS (70 to 80 v 90 to 100), high LDH, low LDH, low albumin, time from primary tumor diagnosis to enrollment, and breast cancer. Impaired pegboard test scores, male sex, more than two brain metastases, low KPS, primary tumor other than breast cancer, high LDH, low albumin, and intermediate time to primary tumor diagnosis were associated with shorter survival (Table 6). DISCUSSION This study used several measurements of NCF and neurologic progression to determine the potential benefit of MGd Table 3. Neurocognitive Function Correlation With Indicator Lesion Volume and Number of Brain Metastases at Baseline Correlation Recall Recognition Delayed Recall Verbal Fluency: COWA Pegboard: Dominant Hand Pegboard: Nondominant Hand Executive Function: Trail B Correlation with indicator lesion volume r P No. of patients Correlation with the number of brain metastases r P No. of patients Abbreviation: COWA, controlled oral word association. Spearman correlation coefficients for sum of indicator lesion volume and z scores of neurocognitive tests. Spearman correlation coefficients for number of indicator lesions and z scores of neurocognitive tests

6 Meyers et al Table 4. Time to Neurocognitive Progression in Lung Cancer Patients Neurocognitive Test Hazard Ratio P, 3 tests Executive, 3 tests Fine motor, 2 tests and executive tests combined, 6 tests All tests combined, 8 tests NOTE. Time to neurocognitive progression was analyzed using SUDAAN software (Research Triangle Institute, Research Triangle Park, NC) by a repeated measures time to event analysis with the tests in each domain and with the three levels of progression included as repeated measures. Abbreviations: MGd, motexafin gadolinium; COWA, controlled oral word association. Hazard ratio 1 favors the MGd treatment arm. Hopkins Verbal Learning Tests recall, recognition, and delay. Trailmaking A and B, and COWA. Grooved pegboard dominant and nondominant. when added to WBRT to address the limitations of a survival end point in assessing the value of a treatment aimed at improving tumor control in the brain. The neurologic progression end point provided a clinical measure of the impact of brain metastases progression on neurologic function. NCF was assessed because it is objective and not dependent on self-reporting, which could be unreliable because of impairment caused by tumor growth and treatment. Although never previously measured in patients with brain metastases, in glioblastoma multiforme, neurocognitive progression is predictive for survival [32]. Performance on neurocognitive tests is related to the patient s ability to manage finances, recognize safe and unsafe behaviors, and remember and comply with medication regimens [21,33]. This is one of the first reports of prospective neurocognitive testing and analysis in patients with brain metastases, including baseline pretreatment assessments. Nearly all patients had some degree of neurocognitive impairment at baseline and multiple neurocognitive impairments were common. NCF at baseline was highly correlated with the volume of the indicator lesions at baseline but not statistically correlated with the number of brain metastases. This suggests that NCF is more affected by tumor burden than by number of lesions. This is a reasonable finding considering these mass effects occur within a fixed intracranial space. NCF at baseline was predictive of overall survival duration in patients with brain metastases. Neurocognitive tests at baseline that were predictive of survival in univariate analysis included memory, motor speed and dexterity, executive function, and global neurocognitive impairment. Only memory was predictive of survival in the multivariate analysis. Pegboard dominant hand performance at baseline was also found to be predictive of survival in a stepwise multivariate analysis that included clinical as well as neurocognitive variables. The finding that memory performance was significant in the model with only neurocognitive tests but not in the model that included clinical variables reflects a statistical dependence between memory performance and the included clinical variables. This suggests that memory function, tumor biology, extent of spread, prior therapy, or other clinical features have an overlapping effect on survival. These neurocognitive results are consistent with those of Mehta et al [20] who reported that test scores of executive function and fine motor coordination bilaterally are significantly associated with survival. Meyers et al [32] have sim- Fig 1. Time to progression (4.5 standard deviation change) in recall memory for lung cancer patients treated with whole-brain radiation therapy (WBRT) with or without addition of motexafin gadolinium (MGd). Patients treated with MGd and WBRT had prolonged time to progression in memory function compared with patients treated with WBRT alone. Median time for MGd is not reached; WBRT is months (95% CI, 9.73 to unknown). Long-rank P.026. D, days; M, months. 162 JOURNAL OF CLINICAL ONCOLOGY

7 Motexafin Gadolinium in Brain Metastases Fig 2. Time to progression (4.5 standard deviation change) in controlled oral word association for lung cancer patients treated with whole-brain radiation therapy (WBRT) with or without addition of motexafin gadolinium (MGd). Patients treated with MGd and WBRT had prolonged time to progression in executive function compared with patients treated with WBRT alone. Median time for MGd is not reached; WBRT is months (95% CI, 9.73 to unknown). Long-rank P D, days; M, months. ilarly reported that cognitive function, specifically performance on a test of verbal memory, is an important prognostic factor in predicting survival in patients with recurrent malignant glioma. The relation between cognitive functioning and survival, observed in these different studies, suggests that cognitive tests are a relatively sensitive measure of the functioning of the brain and that a combination of tumor prognostic variables and brain function assessments seem to predict survival better than tumor variables alone in patients with brain metastases. Neurocognitive testing revealed a benefit in prolonging time to neurocognitive progression in six tests of memory and executive function for patients with brain metastases from lung cancer (but not from breast or other tumor types) treated with MGd and WBRT compared with WBRT alone. A significant delay in time to neurologic progression for patients treated with the MGd and WBRT as assessed by the investigators (P.018, unadjusted) has been found. This benefit was primarily seen in the lung cancer patient group (P.025, unadjusted), as previously reported. In Fig 3. Neurocognitive decline correlates with tumor growth. Median change in neurocognitive test performance (z score) for each neurocognitive test at 2 months versus radiologic response, determined by 2-month magnetic resonance imaging (MRI) response. A positive change in z score indicates a deterioration in the neurocognitive function. COWA, controlled oral word association; NDH, nondominant hand; DH, dominant hand; PD, progressive disease; PR, partial response; Exec Funct, executive function

8 Meyers et al Table 5. Neurocognitive Function Tests As Univariate Predictors of Survival (Cox model) Factor at Baseline HVLT recognition.0011 HVLT recall.0018 HVLT delayed recall.0377 Motor speed and dexterity, pegboard test Dominant hand.0025 Nondominant hand.0148 Executive function Trailmaking Test B.0220 Verbal fluency, COWA.2555 Global neurocognitive impairment, 3 tests impaired.0432 Abbreviations: HVLT, Hopkins Verbal Learning Test; COWA, controlled oral word association. Table 6. Significant Multivariate Predictors of Survival Factor at Baseline Sex.0001 Brain metastases, 1-2 v KPS, v High serum LDH.0001 Low serum LDH.0292 Low serum albumin.0014 Breast cancer.0071 Time from diagnosis to enrollment.0344 Motor speed and dexterity, pegboard dominant hand.0233 NOTE. Multivariate analysis of clinical data and neurocognitive tests by multivariate Cox models. Abbreviations: KPS, Karnofsky performance score; LDH, lactate dehydrogenase. P P that study, a statistically significant benefit with MGd treatment in the prespecified lung cancer subgroup was found by both a blinded review committee and clinical investigators [28]. Although death is a competing end point that is censored in these analyses, the similarity in overall survival between treatment arms minimizes concern that censoring influenced the results. There are several possible reasons why a benefit was observed favoring the MGd and WBRT treatment arm in lung cancer patients for time to both neurocognitive and neurologic progression. Patients with lung cancer in this trial differed substantially from patients with breast and other cancers; lung cancer patients more often presented with brain metastases at their initial primary tumor diagnosis, had brain as the only known site of metastases, had smaller lesion volume, and had less prior therapy [28]. It is likely that less extensive intracranial disease, more rapid and reversible development of CNS signs and symptoms, and less exposure to potentially neurotoxic chemotherapies provide a greater opportunity to demonstrate a benefit in this subgroup. Recently, Sperduto et al [34], in a study of WBRT with or without stereotactic radiosurgery in patients with one to three brain metastases, similarly observed a treatment benefit in a lung cancer subgroup. This study demonstrates that the combination of neurocognitive tests and tumor prognostic variables predicts survival better than tumor variables alone. Addition of MGd to WBRT appeared to improve memory and executive function and prolong time to neurocognitive progression in patients with brain metastases from lung cancer. A randomized phase III trial in patients with brain metastases from NSCLC has been initiated to confirm the benefits seen in this study. Acknowledgment We thank Matt Downs and Heidi Christ-Schmidt at Statistics Collaborate Inc for their invaluable help with the analysis plan, and Gary Koch, PhD, at the Department of Biostatistics, University of North Carolina (Chapel Hill, NC) for his helpful suggestions regarding the SUDAAN analysis. Appendix The appendix is included in the full-text version of this article, available on-line at It is not included in the PDF (via Adobe Acrobat Reader ) version. Authors Disclosures of Potential Conflicts of Interest The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Owns stock (not including shares held through a public mutual fund): J. Smith, Pharmacyclics; S. Phan, Pharmacyclics; R. Miller, Pharmacyclics; M. Renschler, Pharmacyclics. Acted as a consultant within the last 2 years: M. Mehta, Pharmacyclics; W. Curran, Pharmacyclics. Performed contract work within the last 2 years: C. Meyers, Pharmacyclics. Served as an officer of the Board of a company: R. Miller, Pharmacyclics. Received more than $2,000 a year from a company for either of the last 2 years: C. Meyers, Pharmacyclics; M. Mehta, Pharmacyclics. REFERENCES 1. Posner JB: Neurologic Complications of Cancer. Philadelphia, PA: F.A. Davis, Stuschke M, Eberhardt W, Pottgen C, et al: Prophylactic cranial irradiation in locally advanced non-small-cell lung cancer after multimodality treatment: Long-term follow-up and investigations of late neuropsychologic effects. J Clin Oncol 17: , Postmus PE, Haaxma-Reiche H, Smit EF, et al: Treatment of brain metastases of small-cell lung cancer: Comparing teniposide and teniposide with whole-brain radiotherapy A phase III study of the European Organization for the Research and Treatment of Cancer Lung Cancer Cooperative Group. J Clin Oncol 18: , Robnett TJ, Machtay M, Stevenson JP, et al: Factors affecting the risk of brain metastases 164 JOURNAL OF CLINICAL ONCOLOGY

9 Motexafin Gadolinium in Brain Metastases after definitive chemoradiation for locally advanced non small-cell lung carcinoma. J Clin Oncol 19: , Posner JB: Neurologic complication of cancer. Philadelpha, PA, F.A. Davis Company, 1995, pp Wen PY, Loeffler JS: Management of brain metastases. Oncology 13: , Posner JB: Management of brain metastases. Rev Neurol 148: , Sawaya R, Bindal RK: Metastatic brain tumors, in Kaye AH, Laws ER (eds): Brain Tumors. Edinburgh, Scotland, Churchill Livingstone, 1995, p Zimm S, Wampler GL, Stablein D, et al: Intracerebral metastases in solid tumor patients: Natural history and results of treatment. Cancer 48: , Lagerwaard FJ, Levendag PC, Nowak PJ, et al: Identification of prognostic factors in patients with brain metastases: A review of 1292 patients. Int J Radiat Oncol Biol Phys 43: , Nieder C, Nestle U, Motaref B, et al: Prognostic factors in brain metastases: Should patients be selected for aggressive treatment according to recursive partitioning analysis (RPA) classes? Int J Radiat Oncol Biol Phys 46: , 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: , Borgelt 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 6:1-9, Murray KJ, Scott C, Greenberg HM, et al: A randomized phase III study of accelerated hyperfractionation versus standard in patients with unresected brain metastases: A report of the Radiation Therapy Oncology Group (RTOG) Int J Radiat Oncol Biol Phys 39: , Chatani M, Teshima T, Hata K, et al: Prognostic factors in patients with brain metastases from lung carcinoma. Strahlenther Onkol 162: , Chatani M, Matayoshi Y, Masaki N, et al: Radiation therapy for brain metastases from lung carcinoma: Prospective randomized trial according to the level of lactate dehydrogenase. Strahlenther Onkol 170: , Harwood AR, Simson WJ: Radiation therapy of cerebral metastases: A randomized prospective clinical trial. Int J Radiat Oncol Biol Phys 2: , Kurtz JM, Gelber R, Brady LW, et al: The palliation of brain metastases in a favorable patient population: A randomized clinical trial by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 7: , Komarnicky LT, Phillips TL, Martz K, et al: A randomized phase III protocol for the evaluation of misonidazole combined with radiation in the treatment of patients with brain metastases (RTOG-7916). Int J Radiat Oncol Biol Phys 20:53-58, Mehta MP, Shapiro WR, Glantz MJ, et al: Lead-in phase to randomized trial of motexafin gadolinium and whole-brain radiation for patients with brain metastases: Centralized assessment of magnetic resonance imaging, neurocognitive, and neurologic end points. J Clin Oncol 20: , Meyers CA, Boake C: Neurobehavioral disorders experienced by brain tumor patients: Rehabilitation strategies. Cancer Bull 45: , Magda D, Lepp C, Gerasimchuk N, et al: Redox cycling by motexafin gadolinium enhances cellular response to ionizing radiation by forming reactive oxygen species. Int J Radiat Oncol Biol Phys 51: , Miller RA, Woodburn K, Fan Q, et al: In vivo animal studies with gadolinium (III) texaphyrin as a radiation enhancer. Int J Radiat Oncol Biol Phys 45: , Xu S, Zakian K, Thaler H, et al: Effects of motexafin gadolinium on tumor metabolism and radiation sensitivity. Int J Radiat Oncol Biol Phys 49: , Viala J, Vanel D, Meingan P, et al: Phases IB and II multidose trial of gadolinium texaphyrin, a radiation sensitizer detectable at MR imaging: Preliminary results in brain metastases. Radiology 212: , Rosenthal DI, Nurenberg P, Becerra CR, et al: A phase I single-dose trial of gadolinium texaphyrin (Gd-Tex), a tumor selective radiation sensitizer detectable by magnetic resonance imaging. Clin Cancer Res 5: , Carde P, Timmerman R, Mehta MP, et al: Multicenter phase Ib/II trial of the radiation enhancer motexafin gadolinium in patients with brain metastases. J Clin Oncol 19: , Mehta MM, Rodrigus P, Terhaard CHJ, et al: Survival and neurologic outcomes in a randomized trial of motexafin gadolinium and whole brain radiation therapy in brain metastases. J Clin Oncol 21: , Benedict RHB, Schretlen D, Groninger L, et al: Hopkins Verbal Learning Test-Revised: Normative data and analysis of inter-form and test-retest reliability. Clin Neuropsychol 12:43-55, Benton AL, Hamsher KDS: Multilingual Aphasia Examination. Iowa City, IA, AJA Associates, Lezak MD. Neuropsychological Assessment (ed 3). New York, NY, Oxford University Press, Meyers CA, Hess KR, Yung WK, et al: Cognitive function as a predictor of survival in patients with recurrent malignant glioma. J Clin Oncol 18: , Spreen O, Strauss E: A Compendium of Neuropsychological Tests (ed 2). New York, NY, Oxford University Press, Sperduto PW, Scott C, Andrews D, et al: Stereotactic radiosurgery with whole brain radiation therapy improves survival in patients with brain metastases: Report of Radiation Therapy Oncology Group Phase III Study (ASTRO 2002). Int J Radiat Oncol Biol Phys 54:3, 2002 (abstr) 165