Lung cancer is the second most common cancer among men and women and the leading cause of cancer death in both sexes in the United States. In 2003, an estimated 171,900 people in the United States will be diagnosed with lung cancer, and approximately 157,200 will die from the disease.[1] Approximately 35% of lung cancer patients will have locally advanced disease at diagnosis.[2] Central Nervous System Failure Central nervous system (CNS) failure in patients with locally advanced non-small-cell lung cancer (NSCLC) is a common and debilitating problem, with incidence rates ranging between 13% and 54%.[3-12] The risk of CNS failure has been related to disease stage,[11] histology,[13-15] and length of survival from diagnosis.[ 16] Other factors associated with a high risk of CNS failure include female gender,[17] advanced age,[18] preoperative chemotherapy,[13,11] and elevated serum lactate dehydrogenase levels.[17] CNS failure has a profound impact on quality of life and survival; once diagnosed, median survival ranges from 3 to 12 months, depending on Karnofsky performance status (KPS), number of lesions, and status of extracranial disease.[18-20] In most studies, the incidence of CNS metastases is higher in patients with adenocarcinoma and large cell carcinoma.[13-15] Other studies have not found patients with nonsquamous histologies to be at higher risk for CNS failures than patients with squamous cell carcinoma.[5,17] Survival and CNS Metastases
A review of Radiation Therapy Oncology Group (RTOG) data[14] and other combined-modality series[12,13] have shown that chemotherapy significantly decreases extracranial failures but does not decrease the risk of CNS failures. In fact, as survival increases, so does the risk of CNS metastases. Moreover, RTOG data have shown that longer survival among patients with locally advanced NSCLC treated with radiation alone or radiation and chemotherapy is associated with an increased incidence of CNS metastases.[14,16] Recursive partitioning analysis (RPA) of RTOG studies of radiation therapy alone to locoregional disease showed that patients in the two RPA groups with the longest survival had the highest incidence of CNS metastases (18% vs 9%, P = .0004).[16] RPA class I and class II included patients with KPS scores of 80 to 100 and no pleural effusion, distinguishing them from class III and class IV. Class I patients had < 5% weight loss and no lymph node involvement, and received ≥ 66 Gy to the chest. Median survival for class I and II patients was 12.6 and 8.3 months, and for class III and IV patients, 6.2 and 3.3 months, respectively. Recently, several studies have reported excellent median (15 to 25 months) and 2-year (37% to 66%) survival rates with trimodality therapy (chemotherapy, radiation, and surgery) for locally advanced NSCLC.[4,5,12,13,21] These studies also have reported the brain to be one of the most frequent sites of initial failure. Overall, CNS failure rates range from 21% to 54%, with CNS as the first site of relapse in 15% to 30% of patients (Table 1). Prophylactic Cranial Irradiation Randomized Studies
Three randomized trials evaluating prophylactic cranial irradiation (PCI) in patients with locally advanced NSCLC have been published.[7-9] These studies show that PCI decreases or delays the incidence of CNS failure in these patients (Table 2).[4-10,22]

• RTOG Trial-In the early 1980s, RTOG conducted a prospective randomized study comparing PCI (30 Gy in 10 fractions) and chest irradiation to chest irradiation alone in patients with inoperable or unresectable T1-4, N1-3, M0 and resected T1-3, N2-3, M0 nonsquamous NSCLC.[8] Development of symptomatic brain metastases was delayed, but the overall incidence of CNS metastases was not significantly decreased. In a small subgroup of patients with prior complete surgical resection, PCI decreased the incidence of brain metastases from 25% to 0% (P = .06). Most patients in this study did not live long enough to develop CNS failure. Also, the ineffectiveness of locoregional therapy and lack of systemic therapy resulted in a high incidence of locoregional and distant failures that likely were sources of secondary seeding of the CNS after PCI was delivered. Median survival in this study was only 8 months because of ineffective therapy and relatively poor prognostic factors. Median survival in studies reporting the highest rates of CNS failures have ranged from 12 to 25 months.[4,5,12,13,21] According to the RTOG authors: With substantially less than complete efficacy of chest irradiation in controlling lung cancer within the treatment volume, and the absence of effective systemic treatment to prevent dissemination in extrathoracic sites, it is not surprising that PCI produced no measurable impact on survival.[8]
• VA Lung Group Trial-A Veterans Administration Lung Group trial included patients with locally advanced NSCLC who were not candidates for curative resection and who had no evidence of distant metastases.[ 7] Patients were randomized to receive whole-brain irradiation (20 Gy in 10 fractions) or no brain treatment, and one of two regimens of thoracic irradiation. PCI decreased the incidence of brain metastases from 13% to 6% (P = .038) in all non-small-cell histologies and from 29% to 0% in adenocarcinoma (P = .04). There was no difference in median survival.
• M. D. Anderson Trial-Umsawasdi et al from M. D. Anderson Cancer Center treated patients with locally advanced NSCLC with combined chemoradiotherapy and randomized them to PCI (30 Gy in 10 fractions) or no PCI.[9] PCI significantly decreased the incidence of CNS metastases from 27% to 4% (P = .002) and increased the CNS metastasis-free interval. No survival benefit was observed for the treated group.

Nonrandomized Studies
Five nonrandomized multimodality studies in patients with locally advanced NSCLC have demonstrated the potential benefits of PCI (Table 2).[4-6,10,22] In the most notable of these studies, 75 patients with stage IIIA/IIIB NSCLC were treated with induction chemotherapy, preoperative chemoradiotherapy, and surgery. PCI (30 Gy in 15 fractions) was offered to patients accrued during the second half of the study because of a high incidence of brain relapses in the first half. Radiotherapy reduced the rate of brain metastases as the first site of relapse from 30% to 8% at 4 years (P = .005) and the rate of overall brain relapse from 54% to 13% (P < .0001).[5] Skarin et al treated 41 patients with stage III NSCLC with chemotherapy and radiation followed by surgery. Fourteen percent of patients treated with PCI developed CNS metastases, compared to 27% of those who did not receive PCI.[10] The Southwest Oncology Group (SWOG) conducted a phase II study of neutron chest radiotherapy sandwiched between four cycles of chemotherapy in patients with regionally advanced NSCLC.[22] PCI was administered concurrently with chest irradiation (30 Gy in 10 fractions or 36 Gy in 18 fractions). No clinical or radiologic evidence of brain metastases was found in patients who completed PCI therapy. In another phase II SWOG study, patients with stage IIIA NSCLC were treated with chemoradiotherapy and optional PCI (36 Gy in 18 fractions) followed by surgery.[4] Two of 18 (11%) treated with PCI and 24 of 108 (22%) not treated with PCI developed brain metastases. In a phase II Cancer and Leukemia Group B (CALGB) trial, patients with large cell or adenocarcinoma received 30 Gy in 15 fractions to the brain, with neoadjuvant chemoradiation and resection for locally advanced NSCLC. No brain relapse was observed among the 13 patients who received PCI.[6] PCI Toxicity
PCI for patients with locally advanced NSCLC has not become a standard of care despite evidence that it is effective in decreasing CNS failure rates. This is partially due to the lack of a survival advantage associated with the therapy and concern about neurotoxicity. It took several decades for PCI to be accepted as a safe and effective method of managing CNS micrometastases in patients with small-cell lung cancer (SCLC). In this group, PCI has been shown to have a favorable impact on quality of life,[23] decrease the incidence of CNS metastases, and improve survival.[24,25] Neurologic compromise and death is almost certain in the presence of CNS metastases. The benefit of preventing or delaying symptomatic CNS metastases in patients with locally advanced NSCLC, whether or not they are otherwise cured of their disease, cannot be overlooked. PCI toxicity data are derived mainly from studies in SCLC. The highest rates of toxicity have been reported when PCI is given concurrently with chemotherapy or at a high dose per fraction.[2] After low-dose concurrent chemotherapy and PCI and a median follow-up of 6.2 years, 44% of SCLC patients had abnormal neuropsychological tests.[26] In addition, unexpected neurocognitive deficits have been detected in these patients after combination chemotherapy, with no significant change in those deficits after PCI.[27] The authors of this study suggest that the neuropsychological abnormalities associated with SCLC may be secondary to systemic therapy or the disease itself (paraneoplasia). Due to potential paraneoplastic effects and different therapy for NSCLC compared to SCLC, definite conclusions about tolerance of cranial irradiation in NSCLC patients can be drawn only from prospective studies based on serial longitudinal neuropsychological testing in those treated with or without PCI. Late cognitive deficits with the use of PCI for patients with NSCLC have not been detected, partially due to the lack of intensive neuropsychological testing and limited survival. Stuschke et al studied neuropsychological function and brain magnetic resonance imaging (MRI) in patients with locally advanced NSCLC after PCI.[5] T2-weighted MRI revealed white matter abnormalities of higher grades in patients who received PCI than in those who did not. Two of the nine patients treated with PCI and none of the four patients not treated with PCI had grade 4 of 4 white matter abnormalities (diffuse white matter hyperintensity extending from the ventricular lining to the corticomedullary junction). The investigators noted a trend toward impaired neuropsychological functioning among patients with higher-degree white matter abnormalities. Impairments in attention and visual memory in longterm survivors were seen in both PCI and non-PCI-treated patient groups after multimodality therapy. Ongoing Phase III Study
RTOG recently opened a phase III study powered to evaluate the impact of PCI on survival in patients with locally advanced NSCLC. This study is endorsed by all of the North American cooperative groups. Patients who have completed definitive therapy for locally advanced NSCLC and have no evidence of progressive disease or distant metastases are eligible. Patients are randomized to PCI (2 Gy per fraction to 30 Gy) or observation and stratified by histology (squamous vs nonsquamous), therapy (surgery vs no surgery), and disease stage (IIIA vs IIIB). Both groups undergo neuropsychological and quality-of-life testing at regular intervals. It is important that this study be completed now, before strong biases prevent accrual to such a trial. A phase III study is necessary to prove that PCI improves survival by safely preventing CNS metastases in patients who have undergone effective locoregional treatment for locally advanced NSCLC. The successful prevention of CNS metastases will improve quality of life, and for patients controlled extracranially, will improve survival.