Brain metastases represent a frequent problem in several malignancies. They can shorten survival while causing significant morbidity and impairment in the patient's quality of life. Prophylactic cranial irradiation (PCI) has become an integral part of the standard of care in small cell lung cancer (SCLC), yet its role in other malignancies remains the subject of significant discussion. Its role has been extensively investigated in non-small cell lung cancer and less so for breast cancer and other malignancies. Improvements in medical care as well as in whole brain radiotherapy (WBRT) techniques may improve the risk-benefit ratio of this therapy so as to expand its role in cancer care. The use of memantine in WBRT patients as well as the use of hippocampal avoidance techniques are of particular interest in this effort. Herein, we review the history of PCI, its current use, and areas of investigation in the application of PCI.
KEYWORDS : brain metastasis, non-small-cell lung cancer, prophylactic cranial irradiation, small cell lung cancer, whole brain radiation therapy
Prophylactic cranial irradiation (PCI) draws historically from experience in the treatment of the CNS as a sanctuary site of childhood leukemia.
Small cell lung cancer (SCLC) represented a logical leap to a disease well known to frequently and rapidly metastasize to the CNS.
PCI has proven to improve both brain metastasis rates as well as overall survival in SCLC.PCI reduces brain metastasis in non-small-cell lung cancer, but as yet, has failed to demonstrate an overall survival benefit. Many of the more recent trials in the arena have failed to accrue or directly address this issue.
As whole brain radiotherapy (WBRT) techniques and medical management improve, the role for PCI may expand.
Hippocampal avoidance WBRT as well as memantine use in the WBRT setting represent promising techniques for reducing toxicity associated with PCI.
PCI's role in other malignancies, such as breast cancer, remains controversial.CNS involvement remains a common cause of significant morbidity and mortality among a number of malignancies. As such, a number of approaches have been employed in the treatment of this region immediately following diagnosis or early in the course of disease to eradicate micrometastatic deposits, a strategy referred to as 'prophylactic therapy'. Prophylactic cranial irradiation (PCI) arose as a modality for addressing such residual micrometastatic CNS disease in childhood leukemia [1]. The similarities in principle between the treatment of leukemia and that of small cell lung cancer (SCLC) were quickly realized. SCLC, much like leukemia, represents a malignancy that is often widespread at diagnosis with systemic involvement. Treatment of each involves an aggressive chemotherapy regimen to which the disease profoundly, and often completely, responds, and this degree of treatment effect extends to radiotherapy as well. Due to the blood–brain barrier, many systemic therapies fail to reach CNS disease or do so at sublethal levels. Radiotherapy, therefore, represents a particularly suitable methodology for treating micrometastatic disease at this sanctuary site. It should be noted that PCI is actually a misnomer in the sense that its purpose is to treat clinically inapparent disease and not to specifically act as prophylaxis against future metastasis.
The comparison to leukemia led to the consideration and study of the role that PCI may play in lung cancer [2]. The importance of addressing the CNS in SCLC has become more evident as improved imaging has revealed that approximately one in six patients presents with asymptomatic brain involvement [3]. Additionally, the devastating effect upon quality of life and survival that brain metastasis portends raises the stakes of untreated disease of this site. Despite palliation with whole brain radiation (WBRT) among other therapies, up to half of patients still succumb secondary to intracranial progression with the vast majority of patients surviving only 3–7 months [4].
Relapse in the form of brain metastasis continues to be a common occurrence despite advances in therapy. As survival has improved, the high incidence of CNS metastasis in SCLC has become clearer with close to 60% of patients failing in this manner without PCI at 24 months from diagnosis [5]. Autopsy data indicate that the clinical incidence underestimates the actual rates of disease [6]. These rates concur with the extrapolation of clinical trial data as well. Due to these overall very poor outcomes, PCI, though it is certainly not without risks, has been deemed likely to be and has proven to be of significant benefit in well selected patients with SCLC.
More recently, the role of PCI has begun to logically expand to non-small-cell lung cancer (NSCLC). Brain metastasis remains a common outcome of NSCLC with approximately one-third of those receiving therapy eventually being diagnosed with brain lesions [7,8]. The occurrence of this is very early, with 22.5% diagnosed during treatment and 46.5% in the first 4 months from diagnosis [9]. Consequently, clinical trials of PCI in NSCLC have gained increasing acceptance and interest.
In this review, we intend to address the current role of PCI especially in the setting of lung cancer. However, we also will touch upon the broadening role that this therapy may have going forward. Advances in technique such as hippocampal sparing WBRT may improve the risk-benefit ratio of PCI and thus aid in the expansion of its use. Ultimately, we will propose future avenues in other malignancies where PCI is being considered.
Not only is the incidence of brain metastasis at time of diagnosis high among patients with SCLC, but the risk of CNS relapse over time is also very high. It has been estimated that without PCI, intracranial failure occurs in 67% of patients by 2 years [10]. This includes 45% of patients who have brain metastasis as their first and only site of relapse [10]. Initially, with the assumption that reducing these devastating rates of CNS involvement would show clear benefit, multiple randomized trials studied PCI in all-comers regardless of the state of their primary or extracranial disease. This approach was intuitive considering the significant morbidity and high-percentage cause of death that brain metastasis carries. Such trials predominated in the 1980s, but while demonstrating a significant benefit in the reduction in brain metastasis, they failed to show a clear survival benefit [11]. These findings have rightfully been attributed to the heterogeneity of the studied population. Without controlling for the response to initial therapy, the survival of patients could easily have been affected more heavily by competing causes of death from systemic progression of disease.
By the early 1980s, it became clear that complete responders to initial therapy might demonstrate a survival benefit from PCI [12]. Pursuant to this line of thinking, six published randomized trials, summarized in Table 1 , followed to evaluate PCI specifically in complete responders [10,13–17]. Each of these trials showed a clear reduction in brain metastases, but survival differences were too small to statistically detect in these underpowered trials. Auperin et al. conducted a composite meta-analysis including nearly 1000 patients, demonstrating statistically significant absolute overall survival benefit of 5.4 and 8.8% at 3-year follow-up in limited and extensive stage disease, respectively; further, PCI halved the incidence of CNS relapse [18]. This finding in patients with a complete response fits well with the clinician's goals of therapy. If the patient's overall disease is poorly controlled, there is clearly little benefit to subjecting the patient to the side effect profile associated with PCI. Included in this meta-analysis were primarily patients with limited stage disease though extensive stage patients were also included. These results have subsequently been reproduced in both retrospective Surveillance, Epidemiology, and End Results (SEER) Program data and further meta-analysis [19,20]. This more extensive meta-analysis, including over 1500 patients, demonstrated a statistically significant overall survival improvement with a hazard ratio of death of 0.82 for the PCI patients and again a halving of brain metastasis incidence [20].
Published randomized trials evaluating the role of prophylactic cranial irradiation in small cell lung cancer.
| Trial | Years | Patients (n) | PCI dose (Gy/# of fractions) | Brain metastasis rate (%) (PCI vs no PCI) | p-value | Survival (PCI vs No PCI) | p-value | Ref. |
|---|---|---|---|---|---|---|---|---|
| UMCC | 1977–1980 | 29 | 30/10 | 0 vs 36 | 0.02 | [13] | ||
| Okayama | 1981–1986 | 46 | 40/20 | 22 vs 52 | Median 21 months vs 15 months | 0.097 | [14] | |
| PCI-85 | 1985–1993 | 300 | 24/8 | 40 vs 67 (2 year rate) | 29 vs 21.5 (2 year) | 0.14 | [10] | |
| UKCCCR-EORTC | 1987–1995 | 314 | Variable | 38 vs 54 (3 year rate) | 0.00004 | 21 vs 11 (3 year) | 0.25 | [16] |
| PCI-88 | 1988–1994 | 211 | Variable | 44 vs 51 (4 year rate) | 0.14 | 22 vs 16 (4 year) | 0.25 | [17] |
| ECOG-RTOG | 1991–1994 | 32 | 25/10 | 24 vs 53 | NS | Median 15.3 months vs 8.8 months | 0.25 | [15] |
| Auperin Meta-analysis | 1977–1995 | 987 | Variable | 33.3 vs 58.6 (3 year rate) | 20.7 vs 15.3 (3 year) | 0.01 | [18] |
PCI: Prophylactic cranial irradiation.
A number of retrospective studies have raised concerns regarding neurotoxicity after PCI, but these studies have been compromised by many factors. Most importantly, they have been challenged by a lack of baseline neuropsychological testing as prospective cognitive studies have shown up to 97% of limited stage SCLC patients have evidence of cognitive dysfunction prior to PCI [21,22]. Two of the largest randomized trials of PCI incorporated prospective neuropsychometric testing, and both found a significant proportion of patients had cognitive dysfunction at baseline. Yet, they found no differences in cognitive performance between the randomization arms of PCI or no PCI [10,16]. However, higher doses of PCI are associated with worse cognitive outcomes as a randomized trial of different radiation schedules found greater rates of cognitive decline in patients treated with 36 Gy compared with patients treated with 25 Gy [23].
This finding was also borne out in the intergroup randomized study which included 720 patients with limited stage SCLC. Due to the concern that the optimal PCI dose for maximum control of brain metastasis with minimum side effects had yet to be established, this effort set out to compare two doses with three fractionation schedules. Patients either received 25 Gy in 10 daily fractions, 36 Gy in 18 daily fractions, or 36 Gy in 24 twice daily fractions. There was no difference in the incidence of brain metastases between the arms though mortality was higher in the high dose group, likely secondary to increased cancer-related mortality [24]. As such, 25 Gy remains the standard of care for PCI delivery.
Gondi et al. recently analyzed RTOG 0212 and 0214, and demonstrated a decline in Hopkins Verbal Learning Test (HVLT) Recall and self-reported cognitive functioning in patients undergoing PCI versus those who were observed, though these effects were not closely correlated [25].
Because the potential for neurotoxicity from a therapy specifically directed at preserving neurologic function, with only a small survival benefit, is a real concern, Lee et al. have developed a useful decision analysis process that further demonstrates the advantages of PCI in limited stage SCLC patients [26]. However, this model also clearly brings into question the role of PCI should survival rates in SCLC change substantially. Approaches that lower neurotoxicity may become all the more important if SCLC survival rates improve, and the approaches discussed below could become relevant in this context.
Auperin et al. noted in their meta-analysis that there was a small subset of patients with extensive stage disease who experienced improvement in overall survival on par with limited stage patients [18]. However, this cohort represented less than one-fifth of those enrolled, and therefore the results did not gain widespread clinical acceptance.
In practice, most patients present with extensive stage disease. In fact, the proportion of patients with extensive stage disease at diagnosis has risen over the last 40 years, likely secondary to improved imaging for initial staging. While systemic therapy has improved short-term survival rates, long-term control of the disease has remained elusive. Despite numerous advances, the 2 year survival rates for extensive stage disease rose from 1.5 to only 4.6% in the period between 1973 and 2000 [27]. As mentioned above, salvage therapy for these patients has limited success.
In 2007, the European Organization for Research and Treatment of Cancer (EORTC) set out to more directly address the role of PCI in extensive stage disease [28]. They enrolled a broader range of patients according to their response to therapy. Instead of a complete response, any response as defined by the participating center made the patient eligible for PCI. Extensive stage was defined similarly to the traditional definition of disease outside a single radiation portal. Supraclavicular disease or pleural effusions were acceptable for inclusion. The primary objective was to identify possible reduction in symptomatic brain metastases. Several fractionation schedules were permitted including 20 Gy in five fractions, 20 Gy in eight fractions, 25 Gy in ten fractions, 24 Gy in 12 fractions, 30 Gy in 12 fractions, or 30 Gy in ten fractions.
The hazard ratio for symptomatic brain metastases for the PCI group, relative to the control arm, for patients with extensive stage disease was a profound 0.27, implying a 73% reduction in risk [28]. This also translated to an improvement in disease-free survival. More impressively, there was also benefit in terms of overall survival, with a hazard ratio for death of 0.68, in favor of the PCI arm [28]. This equated to an actuarial improvement in the 1 year survival rate from 13.3% for the control group to 27.1% for the PCI cohort [28].
Quality of life assessments were performed at up to 9 months out from therapy, and these demonstrated moderate increases in fatigue, hair loss, appetite changes, nausea/vomiting, and leg weakness in the PCI cohort. However, these were quite tolerable compared with the morbidity of brain metastasis [28]. The primary criticism of this study was its lack of mandatory pre-treatment brain MRI in asymptomatic patients which may have led to inclusion of patients who may have harbored asymptomatic brain metastases. If these patients were randomized to the PCI arm, the cranial radiation could in fact have acted in more of a therapeutic manner, whereas the control cohort patients would have not received any effective therapy for this, and hence could have rapidly progressed from unknown asymptomatic brain metastases to 'intracranial failure'. Nevertheless, PCI has become the standard of care in these patients [29].
More recent confirmatory evidence for these results comes from a pooled analysis from the North Central Cancer Treatment Group, who examined patients with extensive and limited disease with at least stable disease following chemotherapy and thoracic radiotherapy. This 318 patient analysis demonstrated improvement in survival (at 1 and 3 years) with limited toxicity using traditional fractionation schedules [30].
Considering the high rate of brain metastasis following conventional chemoradiotherapy, locoregionally advanced NSCLC represents a logical disease in which the role of PCI should be examined. Improvements in therapy for this disease have led to modest increases in survival and local control. The lifetime risk for developing brain metastasis in locally advanced NSCLC is estimated to be more than 40% with upwards of one-third failing first in the brain [31]. PCI has therefore been evaluated in these patients.
There is clear level 1 evidence to suggest that PCI decreases the subsequent incidence of brain metastasis in locoregionally advanced NSCLC, but not survival.
The Veterans Administration Lung Group was the first to test PCI in a randomized manner for NSCLC [32]. This 1981 study, included 281 patients with inoperable NSCLC who were randomized to either 50 Gy in 25 fractions to their lung disease or a 'short course' of 42 Gy in 15 fractions. They were secondarily randomized to PCI with 20 Gy in ten fractions or observation. There was a decrease in the 2-year rate of clinically detected brain metastasis in the NSCLC group from 13 to 6% with a p-value of 0.038 [32]. The limitations of this study included its small size, low dose of PCI, and heterogeneous chemotherapeutic regimens.
In 1984, Umsawasdi et al. reported on a small randomized study of NSCLC patients who had completed chemoradiotherapy with CAP (cyclophosphamide, doxorubicin, and cisplatin) or chemotherapy alone [30]. Patients were randomized to observation or PCI (30 Gy in ten fractions) if they were without clinical evidence of disease following initial therapy. This study did include a small percentage of patients with Stage I and II disease. The rate of CNS metastasis decreased from 27 to 4% (p = 0.02) [33]. With only 97 evaluable patients, this study was not powered to detect a survival advantage.
In a 1991 RTOG study, a clinically meaningful (from 19 to 9%), but non-significant reduction in brain metastasis was noted [34]. A total of 187 patients with Stage II and III NSCLC were included, and a non-significant reduction in survival was realized in the PCI group. In a Southwest Oncology Group (SWOG) study, 254 patients were randomized with 226 evaluable patients, each with inoperable Stage III disease [35]. Following randomization to either definitive radiotherapy or chemoradiotherapy, patients were randomized to PCI or observation. Again there was significant (p = 0.003) reduction in brain metastasis in the PCI group (1%) versus the observation arm (11%).
Recently, the RTOG published a Phase III comparison of PCI to observation in locally advanced (Stage IIIA and IIIB) NSCLC patients [36]. Of 356 patients accrued, 340 were eligible for analysis. Definitive treatment with either surgery or radiation, post-operative radiation, neoadjuvant, adjuvant, and concurrent chemotherapy were allowed on this trial. Patients were randomized to PCI, 30 Gy in 15 fractions or observation. Unfortunately, this study closed well short of its targeted accrual of 1058 patients which would have represented adequate power to demonstrate a 20% overall survival benefit.
The trial failed to demonstrate statistical significance regarding its primary end point of overall survival with a hazard ratio of 1.03 of death in the observation arm versus the PCI arm (p = 0.86) [36]. However, the rate of brain metastasis was 18.0% for the observation arm versus 7.7% in the PCI cohort with a p-value of 0.004; this represents a hazard ratio of 2.35 and an odds ratio of 2.52 [36]. The rate of grade 3 and 4 toxicities was very low, and most of these were acute in nature, and resolved; the authors reported only four patients with grade 3 or higher late toxicities. These randomized controlled trials of PCI for NSCLC are summarized briefly in Table 2 .
Published randomized trials evaluating the role of prophylactic cranial irradiation in non-small-cell lung cancer.
| Trial | Year of publication | Patients (n) | PCI dose (Gy/# of fractions) | Brain metastasis rate (%) (PCI vs no PCI) | p-value | Survival (PCI vs no PCI) | p-value | Ref. |
|---|---|---|---|---|---|---|---|---|
| VALG | 1981 | 281 | 20/10 | 6 vs 13 | 0.038 | Median 8.2 moths vs 9.7 months | 0.5 | [32] |
| MDACC | 1984 | 97 | 30/10 | 4 vs 27 | 0.02 | [33] | ||
| RTOG 8403 | 1991 | 187 | 30/10 | 9 vs 19 | 0.10 | Median 8.4 months vs 8.1 months | NS | [34] |
| SWOG | 1998 | 254 | 37.5/15 or 30/10 | 1 vs 11 | 0.003 | Median 8 months vs 11 months | 0.004 | [35] |
| RTOG 0214 | 2011 | 356 | 30/15 | 7.7 vs 18 (1 year rate) | 0.004 | 75.6 vs 76.9 (1 year) | 0.86 | [36] |
PCI: Prophylactic cranial irradiation.
In the setting of no demonstrable overall survival benefit from PCI for NSCLC, toxicity and side effects stemming from PCI have become the primary argument for forgoing this therapy in favor of observation. These negative effects with an accompanying reduction in the occurrence of brain metastases, however, must be weighed against the symptomatology that accompanies the emergence of overt symptomatic brain metastases when forgoing PCI.
Quality of life and neurocognitive assessments were reported from the recently completed RTOG trial, which utilized the HVLT, the Activity of Daily Living Scale (ADLS), and the Mini-Mental Status Examination (MMSE) to track deterioration of neurocognitive functioning while also collecting data regarding patient-reported decline in cognitive function, fatigue, and global health/quality of life scores [37].
Quality of life was assessed utilizing the EORTC Quality of Life QLQ-C30 Questionnaire (QLQ-C30) and the BN20. Quality of life data were reported at 6 and 12 months, whereas MMSE and HVLT results were reported at 3, 6 and 12 months [37]. There were no statistically significant differences in the QLQ-C30, the BN20, or patient-reported fatigue at either the 6 or 12 month time points [37]. There was a trend towards appreciable difference (unadjusted p = 0.02, p = 0.24) in patient-reported deterioration in cognitive functioning favoring the observation arm (18 vs 35%) at 6 months that became non-significant at 12 months. At 3 months, the MMSE showed 36% deterioration in the PCI group versus 21% in the observation patients (p = 0.04) [37]. This difference, again, failed to hold at the 6 and 12 month time points. Loss of independence based upon ADLS testing failed to reach statistical significance between the two arms as well. HVLT testing showed the greatest degree of deterioration in the PCI arm. At 3, 6 and 12 months, the HVLT demonstrated statistically significant deterioration in recall in the PCI arm [37]. In both arms, there was a decrease in deterioration of recall at 12 months, compared with the 3 month score. The 1-year incidence of patients with deteriorated HVLT scores was 26 versus 7% for the PCI versus observation cohorts, respectively (p-value of 0.03). A similar trend was noted in HVLT delayed recall. Though not statistically significant at 6 months, this still showed deterioration in the PCI arm at 3 and 12 months. At 1 year, 32% of patients in the PCI group experienced HVLT delayed recall deterioration compared with 5% in their observation counterparts. Of note in this study, the low accrual, far short of the intended power, significantly limited the ability to assess the role of newly developed brain metastasis in terms of possible diminution in neurocognitive and quality of life parameters. Additionally, the 1-year rate of brain metastasis was only 18% in this study. It should, therefore, be recognized that PCI does have risks associated with it, but in high risk populations, the risk of brain metastases and associated decline with growing intracranial lesions often outweighs the cognitive risk of PCI.
The suggestion of worsening memory following WBRT is not novel to this trial. DeAngelis et al. suggested that 1.9–5.1% of patients with known brain metastasis treated with WBRT developed severe dementia; however, in their report, this effect was seen solely in patients treated with non-conventional brain radiation schedules and strategies, and in fact severe dementia was not seen in a single patient treated with a conventionally fractionated radiation course [38]. Other studies have focused upon the early effects of cranial irradiation, especially those occurring within the first few months, though the association of these phenomena with long-term effects have not been established [39–41].
When assessing cognitive changes, it is crucial to recognize the contribution and impact of several other variables, as indicated by Gondi et al. [42]. First, patients may have reduced neurocognitive ability prior to diagnosis or treatment which may or may not have been measured. Second, disease progression, intra-cranially or extra-cranially, may negatively influence neurocognitive as well as quality of life scores. Third, ameliorating medical interventions may cloud the picture or tamper with results. Fourth and finally, a lack in standardization of measurement tools makes cross-study comparison very challenging [42].
The categorical conclusions thus far are that PCI decreases the incidence of brain metastases in patients with locoregionally advanced NSCLC, but this reduction comes with some deterioration in memory as measured by the HVLT test. Therefore, in the future, it would be prudent to evaluate the role of PCI only in NSCLC patients at high risk for developing brain metastases, rather than the general population of all locoregionally advanced NSCLC patients. Several prognostic factors have been proposed as possible stratification variables for such an approach. These include histology, length of survival, age, and stage of disease [43].
Patients with adenocarcinoma have a higher rate of failure in the brain in some series [8]. Andre et al. showed, in a group including patients with cN2 disease, an almost three fold increase in isolated brain metastasis among those with adenocarcinoma versus those with squamous histology. The RTOG database also demonstrates a doubling of brain metastasis in adenocarcinoma versus non-adenocarcinoma, regardless of the use of chemotherapy [44].
As a patient lives longer, his or her risk of brain metastasis rises. Recursive partitioning analysis performed in over 1500 patients in the RTOG database separated patients into four classes based upon multiple prognostic factors [45]. Despite longer lifespan, patients in Class I and Class II exhibited approximately twice the rate of brain metastasis, 18 versus 9% [45]. With improved therapies and multimodality approaches, patients with locally advanced disease have especially benefited from a survival standpoint, leaving them at higher risk of failure in common sites, such as the brain [46].
Age at diagnosis and stage of disease may also be important factors. A Canadian single institution retrospective review of 230 Stage IIIA and B patients receiving bi- or tri-modality therapy reported an overall brain metastasis rate of 34.2% by 2 years [47]. Among this group, age less than 60 was a statistically significant predictor of first failure in the brain with rates of 25.6 versus 11.4% for the less than versus greater than 60 years old groups. The underlying explanation for this phenomenon may be that older patients succumb to intercurrent disease earlier, and do not live long enough to develop brain metastases [47,48]. However, there may also be a difference in the biology of this disease as a function of age. Stage also alters the risk of developing brain metastasis. However, there is some debate as to whether nodal disease (N2) or extent of primary disease (T4) is more predictive of CNS involvement [48,49]. Also, neoadjuvant chemotherapy in trimodailty approaches has been suggested to have a deleterious effect in increasing CNS involvement possibly by delaying definitive treatment of the primary disease [8].
Reparia et al. expanded upon these prognostic factors especially in the cohort of patients with adenocarcinoma [50]. Patients with brain metastases, in this retrospective study, had larger primary disease (p = 0.006) and were slightly older (p = 0.044) than those without. EGFR (25%) and KRAS (28%) were relatively common among patients with brain metastasis. Also, those with EGFR mutations and brain metastasis demonstrated a trend towards increased survival (p = 0.067) versus those with wild type or KRAS mutation disease [50].
In spite of a clear reduction in the risk of developing brain metastases, PCI is currently not part of the standard of care for locally advanced NSCLC patients. In fact, although the absolute decrease may be less dramatic, the relative reduction in brain metastasis with PCI in NSCLC is on par with that of PCI in SCLC that showed a survival benefit in meta-analysis ( Figure 1 ). Concerns continue regarding its long term toxicity especially late neurocognitive effects. The role of PCI should and likely will continue to be investigated, especially in better selected subsets of patients with greater risk of developing brain metastases. Some have suggested that, in such subgroups, instead of PCI, observation with frequent brain imaging in the post-treatment period would identify patients who proceed to develop brain metastases and these patients could be treated while still asymptomatic, as a consequence of 'earlier detection', and that such an approach would improve outcomes, as the brain metastases would be treated 'earlier [51]. However, with median survival close to four months after whole brain radiotherapy, irrespective of whether these are symptomatic or asymptomatic, it is unclear that such a strategy would be widely beneficial [52]. Instead, the hope has remained that a proactive approach utilizing PCI techniques with a reduced risk of cognitive deficits would at least decrease the rate of symptomatic brain metastases, and possibly improve survival in those with high risk NSCLC.