Whole-Brain Radiotherapy in the Management of Brain Metastasis
Deepak Khuntia, Paul Brown, Jing Li, and Minesh P. Mehta  From the University of Wisconsin, Madison,
JCO Mar 10 2006: 1295–1304.

In patients who have brain metastases that impinge on eloquent areas, or are too large, numerous, or disseminated for surgery or RS, Whole-Brain Radiotherapy (WBRT) alone remains the treatment of choice and provides effective symptom relief in the majority of cases. Although response rates after WBRT vary, complete or partial responses have been documented in more than 60% of patients in randomized controlled studies conducted by the RTOG. Table summarizes results of different dose and fractionation schedules from eight randomized studies in patients with brain metastases who received WBRT alone, with median survival ranges from 2.4 to 4.8 months. The consensus from these studies of fractionation schedules is that differences in dose, timing, and fractionation have not significantly altered the median survival time for WBRT treatment of brain metastases. Currently, the most common WBRT regimen uses 30 Gy in ten 3 Gy fractions. Though this dose is clearly inadequate for long-term tumor control, except in the most radiosensitive histologies such as germinoma and lymphoma, this lower dose has traditionally been used to minimize the toxicity of WBRT. An analysis of RTOG trials shows that this results in control of disease in approximately 50% of patients at 6 months.

A second major use of WBRT is as an adjuvant after surgery or radiosurgery. MRI reveals that approximately 80% of patients have more than one metastasis, and approximately 50% have three or more metastases. Moreover, 70% of patients with brain metastases experience relapse after resection, if WBRT is omitted.

CONTROVERSY SURROUNDING THE USE OF WBRT WITH RADIOSURGERY

Radiosurgery after WBRT has been validated with level 1 evidence as a standard of care option in the management of patients with single brain metastases.  As radiosurgery has increased in popularity, a new trend has been emerging in the management of patients with brain metastases; in this approach, patients with a limited number of brain metastatic lesions (the exact definition of limited is based on institutional preference and varies from three to ten or more) are treated with radiosurgery without WBRT, and are then closely monitored, which involves monthly or every other month MRIs. Repeat radiosurgery, is performed for new intracranial metastases, with the intent being avoidance or delay of WBRT in as many patients, and for as long as possible. The putative rationale for this is the avoidance of significant neurotoxicity from WBRT; unfortunately, this approach has not been validated in controlled clinical trials, and its application dramatically increases the overall cost of managing these patients, with multiple and expensive imaging studies (total charges of easily $2,000 per study or more) and repeat radiosurgical procedures (often $20,000 to $40,000 total charge per procedure).

In this section, we will discuss the elements that drive this debate. We summarize the arguments as follows: (1) most patients have oligometastatic disease; (2) survival is unaltered whether upfront WBRT is used or not; (3) radiosurgery is good enough for local control; and (4) the neurologic status and quality of life of patients in whom WBRT is withheld is superior.

A proposition in support of withholding WBRT is that most patients have oligometastatic disease. (The term 'oligometastasis' was initially used to describe a restricted locoregional tumour load, but the term has now become synonymous with isolated distant metastases.)  Older autopsy and computed tomography (CT) imaging studies suggest that the rate of multiple brain metastases ranges from 58% to 86%, with a mean of 66%, but these studies have been criticized on several grounds. Few prospective brain metastases trials have mandated routine MRIs, and therefore these data are sparse. In one MRI-based study, only 19% of the 336 patients had a single lesion; the percentage of patients with two, three, four, and five or more lesions was 16%, 13%, 10% and 40%, respectively. In most trials of radiosurgery, three is considered the upper limit in terms of the definition of oligometastases, and in this trial, 50% of patients had more than three lesions on their MRIs.  Therefore, most reports would suggest that only approximately 20% of patients have one brain metastasis, and this is especially important, as evidence-based data suggest no survival benefit from aggressive local treatments, such as surgery or radiosurgery in patients with more than one metastatic lesion.

A second contention is that survival is unaltered whether upfront WBRT is used or not. In a disease process where the occurrence of brain metastases represents only one component of systemic spread, it is unlikely that local control of disease in one compartment will alter overall survival, and decisions regarding local disease control are not driven by the impact on survival, but rather on the value of local control. Not surprisingly, trials (that were not designed to answer an overall survival question) of local therapies that have excluded WBRT show no difference in overall survival.  However, a cautionary observation is provided by a retrospective analysis from Germany; Pirzkall et al reported that for brain metastases patients without extracranial disease, (ie, patients with a much lower likelihood of dying from systemic metastases) the median survival after radiosurgery alone with WBRT used for salvage was 8.3 months compared with 15.4 months for similar patients treated up front with radiosurgery plus WBRT. Similar results were seen in a retrospective study from the Mayo Clinic with a survival benefit for adjuvant WBRT limited to patients without systemic disease—5-year survival rates of 21% for those who received adjuvant WBRT compared to 4% for those patients who did not.  These observations are crucial, implying that for those patients where prolonged survival is likely, failure to control the intracranial disease by omitting or delaying WBRT could potentially result in a negative survival impact.

It has been proposed, primarily based on retrospective institutional chart reviews, that radiosurgery is good enough for local control. In the prospective randomized Japanese trial, JROSG 99-1, patients were randomly assigned to radiosurgery alone, versus WBRT and radiosurgery. The actuarial 6 month freedom from new brain metastases was 48% in the radiosurgery alone arm, and 82% in the radiosurgery and WBRT arm (logrank, P = .003). Actuarial 1 year brain tumor control rate for the lesions treated with radiosurgery was 70% in the radiosurgery alone arm and 86% in the radiosurgery and WBRT arm (log-rank, P = .019).  Another randomized trial compared radiosurgery alone, WBRT alone, or WBRT and radiosurgery. The local brain control rate was highest in the radiosurgery plus WBRT arm. A prospective single arm, multi-institutional ECOG phase II study of radiosurgery alone for radioresistant histologies (melanoma, sarcoma, renal cell carcinoma) in patients with one to three brain metastases has recently been reported.  At 6 months, 39.2% failed within the radiosurgery volume and 39.4% failed outside the radiosurgery volume, thereby defining the limited benefit from radiosurgery alone. Clinical trial-based assessments therefore suggest high rates of intracranial failures and reduced local control rates when WBRT is omitted or delayed.

It has been speculated that the neurologic status and quality of life of patients in whom WBRT is withheld is superior. The only randomized data available in this context are from a recent Japanese trial, not yet fully published. In that study, patients were randomly assigned to radiosurgery alone or with WBRT; detailed neurocognitive assessments were not performed, and the primary assessment was by an evaluation of performance score and neurologic functional status using RTOG criteria. There were no differences in these end points between the two study arms, belying the claims of worse neurologic outcomes in the WBRT arm.  In fact, many have argued that the converse might be true, (ie, withholding WBRT increases intracranial failure and neurologic deterioration is more directly related to disease progression in the brain.) In a recent phase III trial of WBRT with or without the radiosensitizer, MGd, the most significant predictor for neurologic and neurocognitive decline, as well as deterioration in quality of life was disease progression in the brain.

Therefore, the switch to omitting WBRT has largely been made in the absence of prospective randomized data and needs to be considered carefully. In large measure, it is fair to say that this switch has been made because physicians have observed some patients experiencing neurocognitive and neurologic decline (the causes for which could in fact be multifactorial).

NEUROCOGNITIVE AND NEUROLOGIC DECLINE

Neurocognitive Function: Baseline, Post-WBRT, and Predictive Variables.
Patients with brain metastasis may suffer a certain degree of neurocognitive function (NCF) impairment from multiple factors including the tumor, WBRT, neurosurgical procedures, chemotherapy and other neurotoxic therapies (including anticonvulsants and steroids), or from paraneoplastic effects induced by the malignancy.  Furthermore, radiotherapy variables (eg, total dose, volume of brain irradiated, fraction size) and the interaction with other treatment (eg, concurrent chemotherapy) or patient variables (ie, age, diabetes mellitus) all influence the incidence of radiation-induced injury to the brain and may account for the differences in reported incidences of cognitive deficits.  For example, two studies have not found significant cognitive decline if whole brain fraction size was less than 3 Gy.  In addition, in a phase III clinical trial that compared WBRT (30 Gy in 10 fraction) versus WBRT plus MGd in patients with brain metastasis, analysis of NCF data demonstrated that 90.5% of patients had significant impairment (>1.5 SDs from the age-adjusted population normalized score) in one or more neurocognitive domain at the time of diagnosis of brain metastasis, with 42% of the patients having impairment in at least four out of the eight tests. This result is consistent with previous reports where baseline neurocognitive and neuropsychological test scores were impaired in patients with small-cell lung cancer (even in the absence of overt brain metastases) before receiving WBRT.

On reviewing NCF data from patients with brain metastases who received WBRT alone, we found that at 3 months, the mean NCF test scores appear to be unchanged or slightly reduced. However, 3 months later, gradual recovery is observed . To determine if this recovery is due to patient selection effect, we compared groups of patients who were alive at 4 months after treatment with patients who were alive at 15 months after treatment. We found that 4-month survivors had a sharp drop in their mean NCF scores in the first few months, whereas 15-month survivors had stable NCF with gradual improvement over time. Although both groups of patients had reduction in mean tumor volume from WBRT in the first 4 months, greater reduction was observed in the long-term survivors, which may explain their relatively stable NCF (J. Li, manuscript in preparation). These findings are consistent with a previous report that NCF decline correlates with tumor growth. In addition, these results suggest that the initial changes in patients' NCF test scores, as well as radiologic evidence of tumor regression, may be predictive of overall survival. Although it is possible that the improvement in NCF scores in long-term survivors may also result from increased familiarity with retesting, the battery of NCF tests utilized in this study was designed to have minimal effect from repetitive testing.

A more definitive predictor of survival in patients with brain metastasis is baseline NCF. Univariate analysis has shown that all three domains of NCF testing (memory, fine motor, executive function) at baseline are predictive of survival, whereas multivariate analysis showed that only memory score was an independent predictor.  Similar results were obtained in patients with recurrent brain tumors, where memory test score was found to be highly related to survival by multivariate analysis.  Interestingly in this patient population, cognitive deterioration occurred almost 6 weeks before radiographic failure, highlighting the sensitivity and survival predictive value of NCF changes.  Our preliminary analysis of patients with brain metastasis treated with WBRT has also shown that certain functions may deteriorate earlier in the course of the disease than others. If confirmed, these results may help to further streamline NCF tests that are more sensitive in predicting subsequent QOL changes and survival.

Studies involving NCF deterioration should be interpreted cautiously. NCF decline in the literature is often defined statistically and there is little consensus as to the actual clinical relevance of a statistical definition. Use of different definitions may artificially change the sequence of events, suggesting a further need for a statistical model that allows use of early test scores to predict later events without requiring such predefined statistical windows. In addition, conventionally used measures such as the Folstein mini-mental examination (MMSE) are rather crude and it is crucial to develop sensitive and practical NCF testing to characterize these changes. In particular, the sensitivity of MMSE has been shown to be problematic in detecting subtle neurocognitive dysfunction  in patients with brain metastasis where clinically apparent WBRT-induced dementia is rare (1.9 to 5.1%).  Recent evidence indicates that a battery of validated, language-specific, and population-normalized NCF tests evaluating memory, fine motor coordination, and executive functions confers more accurate and comprehensive measurement of NCF changes in patients with brain metastasis.  

Correlation Between NCF and Quality of Life
Although neurocognitive function is believed to have a major impact on the quality of life (QOL) in patients with many medical, mental, or psychiatric diseases  this relationship has not been adequately evaluated in patients with brain metastases. An early study in patients with primary brain tumors did show that NCF impairment caused a decline in functional independence more often than physical disability. This suggests that any treatment that would reduce the severity of or prolong the time to NCF deterioration could lead to increased functional status in these patients.  The recent development of brief, comprehensive, and validated measurements for NCF and cancer-specific QOL has made it possible to explore the nature of the relationship between NCF and functional performance in anticancer clinical trials.

In patients with brain metastasis who received WBRT alone,  our analysis has shown that at baseline each individual NCF test score is correlated with functional measures such as FACT-Br (Functional Assessment of Cancer Therapy–Brain, brain-specific QOL), Barthel Index, or KPS. The strength of these correlations ranges from weak to moderate, but all are statistically significant. Interestingly, the correlations became stronger 4 months after treatment, and the highest correlation coefficient is observed between memory and Barthel Index. The correlations remain strong at 6 months although statistical significance diminishes due to patient loss (J. Li, manuscript in preparation). These results clearly demonstrate that there is an association between NCF and QOL in patients with brain metastasis. In addition, these results suggest that therapy that preserves patients' NCF, especially memory, may have a positive impact on patients' quality of life.

The sensitivity of the Barthel index and FACT-Br for detecting decline in functional status in patients with brain metastasis or primary brain tumors has been questioned. In fact, it appears from the literature, that gross decline in QOL, especially in terms of severe impairment in activities of daily living may indeed be a late event. Even with the relative insensitivity of current QOL tools, in patients with brain metastasis receiving WBRT, we consistently detected a correlation between NCF and QOL, suggesting that more sensitive QOL testing could help potentially delineate changes in patient's functional status at an earlier point in time.

Mechanism of Neurocognitive Dysfunction
The response to radiotherapy has been classically divided into three categories based on the timing of onset of symptoms: acute, subacute, and late.  Acute effects occur during the first few weeks of treatment and are often characterized by drowsiness, headache, nausea, vomiting, and worsening focal deficits. Often, cerebral edema is the cause of these symptoms and corticosteroids may improve these symptoms. Subacute encephalopathy (early delayed reaction) occurring at 1 to 6 months after completion of radiotherapy may be secondary to diffuse demyelination. Symptoms include headache, somnolence, fatigability, and deterioration of pre-existing deficits that resolve within several months. Late delayed effects appear more than 6 months after radiotherapy and are generally irreversible and progressive. This may be a result of white matter damage due to vascular injury, demyelination, and necrosis. Symptoms range from mild lassitude to significant memory loss and severe dementia  The pathophysiology of radiation-induced neurocognitive damage is complex and involves inter- and intracellular interactions between vasculature and parenchymal cells, particularly oligodendrocytes, which are important for myelination. Oligodendrocyte death occurs either due to direct p53-dependent radiation apoptosis or due to exposure to radiation-induced tumor necrosis factor alpha (TNF{alpha}).  Postradiation injury to the vasculature involves damage to the endothelium leading to platelet aggregation and thrombus formation, followed by abnormal endothelial proliferation and intraluminal collagen deposition. In addition, hippocampal-dependent functions of learning, memory, and spatial information processing seems to be preferentially affected by radiation.  Animal studies reveal that doses as low as 2 Gy can induce apoptosis in the proliferating cells in the hippocampus, leading to decreased repopulative capacity.

Management and Prevention of Neurocognitive Deficits As a Result of WBRT
Treatment (or prophylaxis) of cognitive sequelae of cranial radiation is limited at this time. Methylphenidate has been used in a several small series of patients exhibiting neurobehavioral slowing with limited response.  Patients who develop psychomotor slowing, decline in executive functioning, or general apathy may benefit in particular.  Although these studies suggest beneficial effects with methylphenidate, they have several limitations including small sample size and lack of a blinded control, and therefore the widespread use of methylphenidate should not be considered standard of care.

Erythropoietin has been shown to be a CNS protectant in a number of studies, and this has generated considerable interest in the utilization of this agent.  A blinded, randomized trial of erythropoietin (compared with saline) found significantly less motor impairment in erythropoietin treated rats 2 days after 100 Gy was delivered to the right striatum; by day 10 the erythropoietin treated rats had returned to near control levels while the deficits persisted in the saline-treated rats.   A similar study found that erythropoietin delivered 1 hour after WBRT (17 Gy in one fraction) was neuroprotective in mice.

There has been some interest in using Alzheimer's therapeutic agents to treat radiation-induced injury, since in some aspects radiation-induced injury is clinically and radiographically similar to Alzheimer's dementia. In a trial from Wake Forest University (Winston-Salem, NC), 24 previously irradiated brain tumor patients were treated for 24 weeks with donepezil.  Neurocognitive tests were performed at baseline, 6, 12, 24, and 30 weeks. Verbal fluency, verbal memory, attention, and figural memory scores significantly improved between baseline and week 24, but there was no change on global cognitive function or executive function. No significant worsening of performance was noted on any measures. The limitations of this study include the small sample size and the potential (and uncontrolled) impact of practice (ie, neurocognitive measures repeated over multiple evaluations) and placebo effect.

Prior studies have suggested beneficial effects of vitamin E for patients with Alzheimer's disease.  Researchers at the Queen Elizabeth Hospital in Hong Kong treated 19 patients with temporal lobe radionecrosis with a daily megadose of vitamin E for 1 year, whereas 10 other patients with temporal lobe radionecrosis served as controls (treatment assignment was decided on a voluntary basis).  Significant improvement in global cognitive ability, memory, and executive function occurred among patients in the treatment group after a 1-year medication period. However, as noted by Chan  limitations of this study were that the patients were not randomly assigned or blinded to treatment, and therefore the results should be considered preliminary.

Although a neurocognitive conceptual framework for understanding the effects of radiotherapy is currently limited,  it seems that the pathophysiology of late RT injury is dynamic, complex, and a result of inter- and intracellular interactions between the vasculature and parenchymal compartments, and injury is most likely multifactorial (ie, demyelination, proliferative and degenerative glial reactions, endothelial cell loss, and capillary occlusion).  The vascular hypothesis is probably the most recognized and longest standing premise as the primary cause of radiation-induced damage.  The vascular hypothesis of radiation-induced injury attributes accelerated atherosclerosis and mineralizing microangiopathy that result in vascular insufficiency and infarction to radiation-induced injury and inflammation. Taken together, these mechanisms result in a picture similar to the small vessel disease, as is often seen with vascular dementia.  For this reason there is interest in using pharmaceutical agents that are effective in the treatment of vascular dementia for irradiated brain tumor patients. One of these agents is memantine, a N-methyl-D-aspartate (NMDA) receptor antagonist, that blocks excessive NMDA stimulation that can be induced by ischemia and lead to excitotoxicity. It is believed that agents that block pathologic stimulation of NMDA receptors may protect against further damage in patients with vascular dementia.  Thus, NMDA receptor antagonists such as memantine may be neuroprotective and prevent neuronal injury associated with radiation-induced ischemia. In addition, the physiologic function of the remaining neurons could be restored, resulting in symptomatic improvement.  Preclinical in vitro and in vivo data support this hypothesis.  Phase III clinical trials of memantine in patients with vascular dementia demonstrated clinical benefit, with the subgroup of patients with small-vessel disease responding better to memantine than other types of dementia.  In addition, anecdotal experience using memantine in primary CNS lymphoma patients with cognitive dysfunction after radiotherapy has shown dramatic clinical improvement (I. Robbins, personal communication, October 2005). With the beneficial findings of these studies and the limitations of treatment of cognitive decline after radiotherapy, the RTOG plans a trial of memantine directed at preventing the detrimental effects of cranial radiation.

Besides pharmaceutical interventions, others are considering modifying how WBRT is delivered to decrease the risk of neurotoxicity. As mentioned earlier, doses of 2 Gy or less can damage the hippocampus. As a result, current investigations are underway using new technology to avoid the hippocampus conformally. With the use of intensity modulated radiotherapy, it is possible to create isodose distributions that treat the majority of the brain to full dose, while keeping the radiation dose to the hippocampus relatively low. However, prospective trials with detailed NCF testing will be needed to determine if sparing of the hippocampus alone is beneficial or if other parts of the limbic system will also need to be spared.

In summary, WBRT continues to be a standard and efficacious treatment in the management of brain metastasis. Despite the use of WBRT, outcomes are poor and efforts are being made to incorporate multimodality approaches including surgery, radiosurgery, chemotherapy, and radiotherapy sensitizers to improve survival. Patients with brain metastasis are susceptible to deficits in neurocognition because of their disease and potentially from the treatment for their brain metastasis. Innovative strategies for preventing and treating neurocognitive deficits are actively under investigation.