CHAPTER 250 Fractionated Radiation Therapy for Malignant Brain Tumors
Radiobiology
All malignant neoplasms can potentially be eradicated with suitably high radiation doses. However, because the side effects from unrestricted delivery of radiation to large volumes of normal tissue would be catastrophic, a balance must be struck between treatment efficacy and toxicity, the so-called therapeutic window. Radiation doses are therefore restricted by the tolerance of surrounding tissues, including functional subunits, supporting cytoarchitecture, and vasculature. Radiation tolerance, in turn, is a function of multiple factors, including total dose, dose per fraction, frequency of administration, volume of tissue irradiated, anatomic site, tissue type, comorbid conditions such as hypertension and diabetes, preexisting functional deficits, and the underlying host genetic milieu. External beam RT is delivered through essentially two methods: fractionated external beam RT and stereotactic radiosurgery (SRS). SRS is discussed in detail elsewhere but seeks to limit radiation-related complications by precisely treating a finite volume of disease, usually rather small, with a single treatment. The “radioablative” effect is a combination of tumor cell killing and vascular obliteration engendered by the single high dose of radiation. Fractionation of the radiation dose, in contrast, provides a means of augmenting the dose while attempting to limit detrimental effects on adjacent normal tissue by taking advantage of inherent repair differences between normal and neoplastic tissue.1 Normal tissue exhibits improved repair of sublethal damage, thereby allowing increased cell survival between radiation fractions, which is not the case with most malignancies. The fractionated approach also allows potential reoxygenation of hypoxic regions of the tumor between fractions, thus providing improved radiation efficacy in future treatments through oxygen fixation of radiation damage. Another advantage demonstrated in in vitro systems is that the various phases of the cell cycle express differential radiosensitivity, and fractionated radiation prevents malignant cells from being treated only in a radioresistant portion of the cell cycle, as can happen with single-fraction treatments such as SRS. In the treatment of malignant central nervous system (CNS) entities, typical fraction sizes are between 1.8 and 2.0 Gy daily. A possible disadvantage of fractionation is reduced impact on the vasculature.
Radiation Delivery: Technical Issues
Various advances in radiation technique have improved the therapeutic ratio for fractionated RT by diminishing the volume of normal brain receiving significant radiation doses in patients with primary CNS neoplasms in whom partial brain irradiation is desired. The margin required for daily setup variability has been decreased by patient immobilization devices, such as a molded Aquaplast face mask; such devices enhance interfraction reliability in patient position and constrain intrafraction motion. Improvements in imaging with the advent and institution of cross-sectional imaging have allowed more precise delineation of targets, although a significant margin, particularly with gliomas, remains necessary to account for extension of subclinical disease.2 Most commonly, planning for management of CNS neoplasms is accomplished with the fused product of high-resolution diagnostic magnetic resonance imaging (MRI) and computed tomography (CT) performed in the treatment position (Fig. 250-E1). Further accuracy in localization of treatment is offered by image-guided RT, which permits pretreatment visualization of patient anatomy and tumor location, along with the ability to register this information with baseline readings. This approach allows daily adjustments to ensure accurate delivery of radiation and decreases the target volume by minimizing the range of setup error. Although frequently unnecessary in the treatment of CNS neoplasms, given the relatively static location of intracranial structures when the head is constrained by immobilization devices, daily image-guided RT using biplanar orthogonal x-ray imaging systems, cone beam CT, or megavoltage CT further improves positional reproducibility and permits decreased margins.
Three-Dimensional Planning and Treatment
As alluded to earlier, CT-based planning has catalyzed the evolution of three-dimensional conformal radiation therapy (3DCRT) in which non-coplanar treatment fields with unique entrance and exit pathways can be directed onto a patient-specific anatomic target. This allows the desired dose to be displayed in a volumetric distribution with modern treatment planning software. Avoidance of critical structures proximate to the treatment volume, such as the brainstem, optic apparatus, and spinal cord, can subsequently be enhanced by the precise delineation of volumetric distribution allowed by such planning. Early research confirmed the benefit of three-dimensional treatment planning in terms of the ability to reduce the volume of brain receiving full-dose treatment (95% isodose line) by 30% with nonaxial techniques when compared with conventional parallel opposed orientations.3 Intensity-modulated radiation therapy (IMRT) is a subset of 3DCRT that provides improved conformality through nonuniform delivery of doses by modulating the photon flux of specific beams during treatment. This approach results in highly shaped radiation dose distributions and is especially relevant when it is necessary to produce concave or convex distributions, which is of significance when tumors are in close proximity to the optic apparatus, vestibulocochlear structures, hypothalamic-pituitary axis, hippocampus, brainstem, and other structures. IMRT has generally been restricted in CNS applications because of the adequacy of 3DCRT plans in most situations. In addition, the tight conformality of doses within the high-dose regions provided by IMRT is typically accomplished through the delivery of lower doses of radiation to larger areas. Long-term data on the effect of normal regions of CNS receiving low doses of radiation are inadequate, although some dosimetric evaluations have countered this concern.4
Conventional External Beam Radiotherapy
WBRT is most commonly delivered via parallel opposed lateral fields and 4- to 10-MV photons with an Aquaplast face mask or other custom immobilization (Fig. 250-E2A). Treatment simulations can be performed clinically on a kilovoltage x-ray machine or by CT with digitally reconstructed radiographic projections. To account for day-to-day intratreatment and intertreatment variability in setup and dosimetric factors such as the penumbra (a region of dose falloff at the margin of the radiation field where “cold spots” can occur), a margin of at least 1 cm is advocated. Bony landmarks for the intracranial contents include the calvaria, cribriform plate, and bases of the middle and posterior cranial fossae. Ideally, the lenses are spared from the radiation through appropriate field blocking and beam, couch, and collimator angling; this detail is important because delayed cataract formation can easily occur, even with very modest radiation delivered to the lens (Fig. 250-E2B). WBRT fields cover the brain to the bottom of the C1 vertebra, although additional extension to C2 is designed by convention for small cell lung cancer (SCLC), PCL, or overt disease in the posterior fossa. In addition, the WBRT field for CNS lymphoma, described as a German helmet, is extended anteriorly by marking the fleshy canthi in an attempt to cover the posterior half of the globe given the tendency of CNS lymphoma to involve the orbits. Inadequate attention to bony anatomic landmarks and appropriate margins as defined earlier can lead to regional underdosing, and isolated relapses within the inferior frontal lobes or posterior fossa have been reported. Cataract development and injury to the lacrimal gland appear to be minimized by appropriate design.
Fractionated Stereotactic Radiotherapy
Fractionated stereotactic radiotherapy (FSRT) is a hybrid technique that incorporates the precision of radiosurgery and the improved therapeutic ratio of conventional, fractionated RT. FSRT commonly uses multiple non-coplanar fixed fields and precision planning facilitated by improved patient tracking or immobilization. The advent of robotic devices and mini-multileaf collimators permits very precise and conformal dose delivery with this method, also similar to that offered by SRS (Fig. 250-E3). An additional advantage in the cranium, because of its spheroidal geometry, is access to numerous beam entry points, thereby further improving dose conformality. Various FSRT systems with reported precision between 1 and 3 mm have been developed.5,6 FSRT is used for lesions requiring high precision that cannot be appropriately treated with SRS. Common disqualifications for SRS include a maximum dimension greater than 4.0 cm and location in a critical anatomic region that is radiation sensitive and cannot be avoided by the high-dose region of the SRS dose distribution.
Charged particle radiation is a specific type of external beam radiation that delivers charged particles (most commonly protons) rather than photons. Charged particles have the inherent advantage of deposition of the majority of the dose at a depth that depends on the initial energy, thus avoiding the typical exit dose present in photon therapy. Avoidance of the exit dose can theoretically decrease the integral radiation dose significantly. The region of maximum dose deposition is known as the Bragg peak. Narrow Bragg peaks can be achieved with pencil scanning proton beams, but coverage of larger volumes typically requires the proton beams to be modified by passive range modulators to disperse the Bragg peak and broaden deposition of the dose. Although proton beam therapy can be used in any CNS location, it has been applied more commonly for tumors of the skull base, such as chordomas and chondrosarcomas; these tumors require high doses for durable local control and are surrounded by essential radiosensitive structures, which limits dose escalation when using non-IMRT photon fields.7,8 Proton beam therapy has also been explicitly advocated for the treatment of childhood tumors because of the decreased integral radiation dose.9 There are in fact several limitations with this modality. The number of proton therapy facilities worldwide has been limited because of the immense expense of construction and operation, usually 10- to 100-fold greater than the cost for a conventional facility. Proton therapy facilities are, however, anticipated to become somewhat more common, with several facilities recently being commissioned or in the planning or construction phases. The vast majority of these facilities are built with older cyclotron-based technologies, the inherent limitations of which include significant expense, space needs, technology support, neutron contamination (with the associated concern of increased second malignancies),10 and lack of intensity modulation, which implies that the overall dosimetric advantages over sophisticated IMRT plans are indeed limited.11 Several modern technologic advances will make it possible to eliminate the neutron contamination, miniaturize the equipment so that it can fit in standard hospital facilities, reduce cost by fivefold or greater, and incorporate proton intensity modulation, which will provide genuine dosimetric advantages. There are clearly concerns that as these facilities proliferate, overuse might become commonplace.12 Overuse is especially worrisome because no level 1 evidence has demonstrated the superiority of protons over photons in any clinical setting.
Brain Metastases
The exact incidence of brain metastases remains unknown because brain metastasis as an event is not coded in cancer databases. Estimated rates depend on whether the incidence is calculated from autopsy data, clinical studies, tumor registries, hospital records, or other sources.13 If the rate of brain metastases is calculated from autopsy data, the annual estimated incidence in the United States is as high as 170,000 cases per year.14 Of the patients in whom brain metastases develop, the vast majority die within a few months of diagnosis, thus making brain metastasis one of the most common immediate causes of death in the United States. Lung cancer accounts for approximately half of all secondary tumors to the brain; other primary tumors commonly causing brain metastases include breast cancer and melanoma. Less commonly, primary tumors of the gastrointestinal tract and genitourinary system, lymphomas, sarcomas, and prostate cancer also metastasize to the brain. In terms of disease-specific risk, melanoma has the greatest likelihood of metastasizing to the brain. The overall incidence of brain metastases is probably increasing because of the combination of better diagnostic techniques and small gains in systemic therapy. Improved systemic therapeutic options have altered the conventional disease course such that patients with primary cancers live longer. With longer survival, asymptomatic micrometastatic disease in the brain is more likely to become overt, thereby inflating the incidence of the problem. Furthermore, the brain has traditionally been thought to represent a sanctuary site by not permitting penetration of most cytotoxic chemotherapeutic agents when the blood-brain barrier is intact. Effective eradication of systemic micrometastatic disease therefore introduces the possibility of having brain-only metastatic disease remaining, which is then expressed clinically over time.
Older data from the 1970s suggest that the median survival time of untreated patients with brain metastases is approximately 1 month.15,16 The principal cause of this rapid mortality is uncontrolled peritumoral vasogenic edema leading to herniation and rapid death. As a consequence, measures to decrease vasogenic edema are commonly used in the initial management of patients with metastatic brain tumors, and they provide relief of symptoms in the majority of patients.16 In particular, glucocorticoids have been shown to control vasogenic edema and result in minimal prolongation of median survival to approximately 2 months.15,17
Fractionation Trials
Chao and colleagues first reported the value of external beam RT for brain metastases in 1954.18 Subsequently, the Radiation Therapy Oncology Group (RTOG) conducted a series of studies exploring a variety of fractionation schedules and evaluated their impact on outcome in patients with brain metastases. These trials were primarily conducted in the 1970s and focused on identifying the appropriate schedule and dose.19,20 Table 250-1 summarizes the outcome data from the first three RTOG brain metastasis fractionation trials.
Based on more than 2000 patients enrolled in these three RTOG trials, it was apparent that no specific fractionation schedule was markedly superior. The dose range that was explored was 20 to 50 Gy, with median survival ranging from 15 to 21 weeks. Response rates after WBRT varied, but complete or partial responses were documented in approximately 60% of patients, with the caveat that the imaging was suboptimal by modern standards and the response rates are most likely overestimated.21 The best median survival was seen in a group of 233 patients treated with 30-Gy WBRT delivered over a period of 2 weeks. This cohort of patients exhibited a median survival of 21 weeks. As a consequence, this schedule has become the most commonly used fractionation scheme for patients with brain metastases treated in the United States over the past 2 decades.22 Although hypofractionated schedules are infrequently used in the United States, WBRT delivered over a single week (20 Gy in five fractions) is not uncommon in Europe.23
Dose-Response Relationship
There appears to be a dose-response relationship between radiation dose and local control. This issue is of critical significance in that in the older RTOG studies 50% or more of patients died of neurological deterioration/progression, and it would be logical to assume that if such deterioration could be controlled, survival may be enhanced or at least the quality of survivorship might be improved. To test this hypothesis, the RTOG conducted study 8528 to evaluate the role of dose escalation for solitary brain metastases via accelerated hyperfractionation (1.6 Gy twice daily up to 32 Gy to the whole brain followed by a tumor boost to total doses ranging from 48 to 70.4 Gy). This trial demonstrated a significant advantage in survival and neurological improvement with higher doses, thus suggesting that intracranial disease control is related to dose and that such control actually translates into neurological improvement and a survival advantage.24 However, the confirmatory trial, RTOG 9104, failed to validate this observation.25
The relationship between local control and total dose has been further inferred by case-control analysis. Nieder and colleagues treated a group of 164 patients with the standard WBRT regimen of 30 Gy in 10 fractions, monitored all patients by serial CT imaging, and correlated local control and survival with various prognostic factors.26 Subsequently, 39 patients were treated to a total dose of 40 to 60 Gy and compared with the cohort of 164 patients treated with a total dose of 30 Gy by using the matched-cohort analysis method. The matching procedure produced equivalent groups of patients and showed a significant dose effect, with 30 Gy resulting in a local response rate of 50% versus 77% for doses in the 40- to 60-Gy range. In this analysis, although local control improved from 50% to 77% by escalating the radiation dose (P = .05), survival was not significantly altered. The lack of a distinct survival advantage demonstrated the competing causes of mortality in patients with brain metastases and confirmed that local control will improve survival only in patients who are not experiencing simultaneous systemic failure. In a subsequent analysis from the same group, the CT scans of 322 patients were analyzed to specifically evaluate the impact of dose (in the 25- to 50-Gy range) on local control. The total dose was recalculated with the linear quadratic model and expressed as the biologically effective dose for tumor (BED10). BED10 values ranged from 37.5 to 72 Gy, and the best local control was found to be a function of a higher BED, the number of brain metastases, and the histology of the primary tumor. An increasing BED resulted in a significant decline in 1-year failure rates, with low BED values resulting in failure rates of 44% as opposed to 31% for high BED values. Once again, overall survival was not dependent on the total dose.27 In a large analysis, Lagerwaard and coworkers identified prognostic factors in a group of almost 1300 patients treated for brain metastases.28 In this multivariate analysis the single most important independent prognostic factor predicting improved survival was treatment modality, with more aggressive therapy having the potential to improve survival.
How does one interpret these conflicting data? It is quite clear that a significant proportion of patients with brain metastases succumb to systemic disease, and therefore enhancing control of intracranial diseases is unlikely to provide a survival benefit to this group of patients. In clinical trials, whether prospective or retrospective, in which a significant majority of patients harbor considerable systemic disease that will dictate the outcome, improvement in survival from more aggressive intracranial local control is unlikely to be demonstrated. However, in clinical situations in which patient selection identifies individuals who are less likely to rapidly succumb to systemic progression and patients are at risk for dying of intracranial disease, local control of brain metastases becomes critical. This has best been demonstrated in surgical trials comparing resection and WBRT with WBRT alone. Two such randomized trials by Vecht and associates29,30 and Patchell and colleagues31 have validated this paradigm by showing improved survival in the surgical arms. A third study, by Mintz and coworkers, did not find an improvement in overall survival when surgery was added to WBRT.32 A slightly higher percentage of patients had disseminated systemic disease at the time of enrollment than in the previous two studies, again emphasizing the difficulty of teasing out survival benefit from improved local control of CNS metastases when other risk factors for disease-specific mortality are in play.
Prognostic Factor Analysis
The role of WBRT and, more specifically, escalated delivery of radiation to brain metastases has at times been obscured by the difficulty in identifying patients most likely to succumb to their intracranial disease versus systemic progression. Among the heterogeneous population of patients with brain metastases, those with limited life expectancy because of a heavy systemic burden of disease would obviously derive less benefit from aggressive intervention than would those with prolonged anticipated overall survival. Some of the seminal work in this area of subgroup stratification was done by the RTOG, which identified a set of prognostic criteria predictive of survival. Such criteria included age younger than 60 years, a Karnofsky Performance Scale (KPS) score higher than 70, controlled primary tumor, absence of extracranial metastases, and three or fewer lesions.19,33,34 In a seminal analysis, 1200 patients from three consecutive RTOG trials conducted from 1979 to 1993 were amalgamated into a single database, and a statistical methodology known as recursive partitioning analysis (RPA) was used to identify subsets of patients based on evaluation of prognostic factors. This analysis suggested that patients with brain metastases can be broadly categorized into three classes with different outcomes (Table 250-2). The most favorable group, class I, includes patients with a KPS score of 70 or higher, age younger than 65 years, controlled primary, and absence of extracranial metastases. These patients have a median survival of 7.1 months. Patients in the worst prognostic group, class III, were characterized by a KPS score of less than 70, and their median survival was 2.3 months. The intermediate group, described as class II, included all patients not specifically within class I or III and had a median survival of 4.2 months.35 The discrepancy in anticipated overall survival between RPA classes emphasizes the importance of appropriate patient selection in studies evaluating the impact of aggressive interventions.
Although the RTOG RPA classification has been useful in trial design, additional factors appear to play a role in determining overall survival. Despite not being found to be significant in the RTOG RPA, the number/volume of brain metastases probably has some impact. Patients with four or more brain metastases appear to have a poorer prognosis, but the RPA classification does not categorically identify this factor, in part because the database contained patients imaged with both CT and MRI whereas current practice relies primarily on MRI.36 Similarly, the distinction of single versus multiple brain metastases was not found to be statistically significant in the RTOG RPA system, but analysis of a study by Patchell’s group,37 which evaluated postoperative RT for single, resected brain metastases, showed no difference in overall survival regardless of patient class (I versus II).38 A large, retrospective study of 916 patients treated from 1985 to 2000 within a single institution showed improvement in overall survival in both class I and II when patients with single brain metastases were compared with those with multiple metastases within the same class, again inferring relevance to the number of brain metastases.39 A prognostic score index for radiosurgery in patients with brain metastases (SIR) has also been developed; scores between 0 and 10 are assigned on the basis of lesion number and the largest lesion volume, as well as age, KPS score, and extent of extracranial disease.40 This index has been validated in another smaller patient cohort.41 Finally, primary tumor type probably also has some role in predicting survival.
Most recently, a new prognostic index described as the Grade Prognostic Index (GPA) was derived from an RTOG database of 1960 patients with brain metastases, including the recently reported RTOG 9508 (Table 250-3).42 The four components included as part of the GPA are age, KPS score, extracranial metastases (none and present), and number of metastases (one, two to three, more than three). Each category receives a score of 0, 0.5, or 1, with a cumulative range between 0 and 4. This index overcomes some of the drawbacks of the RPA by incorporating the number of metastases and removing the frequently ambiguous descriptor of a controlled primary site, which could vary depending on the type, timing, and technique of restaging studies. The GPA does not require that treatment planning information, such as the tumor volume required by SIR, be available, thus permitting its use in guiding the choice of treatment. Based on the GPA, median survival times were 2.6 months for a GPA of 0 to 1, 3.8 months for a GPA of 1.5 to 2.5, 6.9 months for a GPA of 3, and 11.0 months for a GPA of 3.5 to 4.0, it and was strongly significant (P < .0001).
Role of Adjuvant Whole-Brain Radiotherapy
Level 1 evidence has shown an improvement in overall survival with surgery for single metastases31 and SRS for solitary metastases and less than level 1 evidence (subset analysis of a prospective randomized trial) for one to three metastases with age younger than 50 years, non–small cell lung carcinoma (NSCLC) primary, and RPA class I.43 With the emergence and increased use of these local control modalities for brain metastases, the role of adjuvant WBRT has been questioned. WBRT has been used in this clinical situation under the rationale that micrometastatic disease in the brain needs to be controlled. A multi-institutional, prospective, randomized trial evaluated the value of postoperative WBRT in patients managed by surgical resection with or without postoperative WBRT (Table 250-4).37,44 Ninety-five patients with single brain metastases were treated by complete resection as verified by postoperative MRI and subsequently randomized to postoperative WBRT, 50.4 Gy in 28 fractions of 1.8 Gy each, or no postoperative WBRT. Forty-nine patients were assigned to the WBRT group and 46 to the observation group. Recurrence of tumor in the brain was significantly less frequent in the RT group (18%) than in the observation group (70%). Patients in the RT group (14%) were also less likely to die of neurological causes than those in the observation group (44%). There was, however, no significant difference in overall survival between the two groups, even though the study was not powered to detect a survival difference (Table 250-4). The lack of impact by WBRT on overall survival may be partially attributable to the high proportion of patients (61%) in the surgery-alone arm who received WBRT at the time of recurrence.45 Similar results were found when evaluating SRS with or without adjuvant WBRT. Aoyama and coauthors published a prospective phase III trial that randomized 132 patients (mostly lung cancer) with a KPS score of 70 or greater and four or fewer metastases to SRS with or without WBRT.44 No improvement in overall survival was observed; median survival time and the 1-year survival rate were 7.5 months and 38.5% in the group treated with WBRT plus SRS and 8.0 months and 28.4% for SRS alone (P = .42). Parenthetically, this study was also not adequately powered to detect improvement in survival. A significant increase in freedom from new brain metastases was identified in patients receiving WBRT (76.4% versus 46.8%, P < .001). These two studies demonstrate that postoperative WBRT significantly diminishes intracranial relapse and neurological death in well-selected patients with single brain metastases undergoing either SRS or resection without having an impact on overall survival.
Those who argue against the use of adjuvant WBRT cite the lack of overall survival benefit with adjuvant WBRT, the possibility of using WBRT or repeat SRS for recurrence, and potential late neurocognitive side effects from WBRT.21 Although no survival benefit was demonstrated in either the Aoyama or Patchell studies, neither was adequately powered for this end point, and in both, a significant proportion of patients underwent salvage WBRT. A retrospective study from the Mayo Clinic examined patients with single brain metastases treated by surgical excision and further stratified by the extent of resection, the presence of extracranial disease, and the use of WBRT. The 46 patients with complete resection and no systemic disease treated with WBRT were found to have improved overall survival (median, 1.3 years; 2-year survival rate, 41%; 5-year survival rate, 21%) when compared with the 75 patients who did not receive adjuvant WBRT (median, 0.7 year; 2-year survival rate, 19%; 5-year survival rate, 4%).46 In a recent retrospective review, local control was found to be enhanced in patients who underwent surgical excision followed by WBRT and a tumor boost (RR 2.31) or complete surgical resection (RR 3.79).47 This improved local control parlayed into an improvement in overall survival.
Radiation Sensitizers
Because of the persistence of local failure despite escalating doses from 20 to 50 Gy, the RTOG strategy in the 1980s focused on exploring the use of radiation sensitizers to improve outcome in this group of patients. Two clinical trials using significantly different sensitizers, misonidazole, a hypoxic cell sensitizer, and bromodeoxyuridine (BUdR), a halogenated pyrimidine and an S-phase sensitizer, were conducted. In a four-arm trial evaluating misonidazole, no significant survival advantage was noted. Similarly, the BUdR trial failed to show a significant survival advantage.48 Recently, the RTOG evaluated the antiangiogenic agent thalidomide49 and the chronomodulator melatonin50 as putative radiosensitizers in the treatment of brain metastases, and neither showed a survival benefit.
More recently, motexafin gadolinium (MGd), a metalloporphyrin redox modulator that has demonstrated selective tumor localization with potentiation of reactive oxygen species and depletion of thiol-based free radical scavengers, was tested in a randomized fashion as an adjunct to RT. No difference in overall survival (5.2 months for MGd versus 4.9 months for WBRT alone, P = .48) or time to neurological progression (9.5 months for MGd versus 8.3 months for WBRT alone, P = .95) was identified, although patients with NSCLC, who represented greater than 50% of the population, had a significant improvement in time to neurological progression (median, not reached for MGd versus 7.4 months for WBRT; P = .048).51 A follow-up phase III trial randomized 554 patients with NSCLC to WBRT with or without MGd. The primary outcome was neurological progression, and a trend toward improvement was found with MGd (15 versus 10 months, P = .12). The number of patients requiring retreatment of intracranial disease also decreased when MGd was administered (P < .001). In a preplanned partition by geographic region, North American patients, in whom treatment was more prompt, were found to have a statistically significant prolongation of time to neurological progression from 8.8 months for WBRT to 24.2 months for WBRT plus MGd (P = .004; hazard ratio [HR], 0.53).51a
Similarly, a phase III trial evaluating WBRT with or without efaproxiral, a synthetic small molecule that alters the oxygen-hemoglobin dissociation curve and thereby increases capillary PO2 and presumably tumor oxygenation, showed no difference in median survival. In the subgroup with NSCLC or breast cancer, median survival trended toward improvement, 6.0 and 4.4 months, respectively (HR, 0.82; P = .07), thus suggesting a potential survival benefit, with most of the benefit being derived from breast cancer patients.52 A phase III trial examining the potential benefit in women with primary breast cancer, the ENRICH trial, recently closed and the results are pending.53
Side Effects of Whole-Brain Radiotherapy
Modern, appropriately fractionated WBRT is generally well tolerated; acute side effects are typically limited and can include fatigue, hair loss, particularly along the midline and vertex, erythema, and otitis. Impairment of neurocognitive function months to years after WBRT is the most significant potential late side effect of radiation and is the most common reason cited for not using RT in the adjuvant setting. DeAngelis and coauthors reviewed the Memorial Sloan-Kettering Cancer Center’s patients from 1978 to 1985 treated postoperatively with WBRT (79 patients) and no RT (19 patients).54 In 4 (11%) of the WBRT patients who survived longer than 1 year, neurotoxicity with progressive dementia, ataxia, and urinary incontinence presumed to be secondary to the radiation developed. All these patients had received at least part of their radiation course at greater than 3 Gy per fraction, a hypofractionated approach that would not fall under the current standard of care in North America. A similar series involving an overlapping patient cohort from the same institution identified a rate of radiation-induced dementia of 1.7% and 5.1% for those receiving WBRT alone or after surgery, respectively.55 Again, the majority of these patients (75%, 9 of 12) had their WBRT delivered at greater than 3 Gy per fraction.
Prospective data evaluating neurocognitive function, however, have shown a more complex picture, with baseline detriment in neurocognitive function created from a confluence of factors, including the presence of brain metastases, neurosurgical interventions, chemotherapy, and other neurotoxic therapies such as steroids and anticonvulsants.56 In a phase III trial evaluating WBRT with or without MGd, 90.5% of patients exhibited impairment in one or more neurocognitive tests at baseline.57 Test scores of memory, fine motor speed, executive function, and global neurocognitive impairment at baseline correlated with brain tumor volume and were predictive of survival. Neurocognitive deterioration was correlated with poor response to WBRT and disease progression.58 In a correlative study involving SRS with or without WBRT, mini-mental status examination (MMSE) scores correlated with KPS score, tumor volume, extent of tumor edema, and age.59 Control of the brain tumor was concluded to be the most important factor in stabilizing neurocognitive function. In an attempt to isolate the effects of WBRT from the presence and potential progression of macroscopic brain metastases, 96 patients with SCLC treated by prophylactic cranial irradiation (PCI) were evaluated by formal neurocognitive testing.60 Again, a large number of patients (47%) had impaired neurocognitive function before WBRT. Sixty-nine patients underwent PCI and were compared with those who deferred PCI in this nonrandomized study. Although significant transient impairment was identified in executive function and language at early time points in patients who underwent PCI, these deficits were not sustained, and no difference was detected between the two groups at late time points. Similar cognitive deficits at baseline but no ultimate detriment in neurocognitive function attributable to PCI have been found by other groups despite the inclusion of large dose fractions of radiation.61
Prophylactic Cranial Irradiation
Several diseases have a natural history that involves frequent metastases to the brain. Given the relatively impervious nature of an intact blood-brain barrier to chemotherapeutics when compared with other body compartments, some patients have had systemic eradication of their disease with a combination of surgery, RT, or chemotherapy only to ultimately have fatal brain parenchymal metastases subsequently develop. PCI was instituted in such populations in an attempt to improve overall survival by eradication of disease from this potential sanctuary site and has been most classically used for limited-stage SCLC. A phase III study evaluating 300 patients with SCLC who had a complete response to systemic treatment and were subsequently randomized to PCI or observation showed a decrease in the brain as a first site of relapse (45% versus 19%, P = 10−6) and in the 2-year rate of development of brain metastases (67% to 40%, P = 10−13), but a nonstatistical difference in overall survival at 2 years (21.5% versus 29%, P = .14).62 No significant difference between the groups was found in terms of neuropsychological function. A subsequent meta-analysis found an improvement in overall survival rates with PCI at 3 years from 15.3% to 20.7%.63 PCI for locally advanced (stage III) NSCLC has also been advocated and studied, albeit less extensively than for SCLC. Unfortunately, a large, randomized trial evaluating PCI for stage III NSCLC closed early because of slow accrual. PCI has classically been confined to limited-stage SCLC, but a recent European Organization for Research and Treatment of Cancer (EORTC) study found a survival advantage in patients with extensive-stage SCLC.64
Reirradiation of Brain Metastases
The use of relatively moderate doses of WBRT frequently results in a clinical situation in which patients have progressive brain metastases but their performance status or size and number of metastases do not qualify them for either resection or SRS. This is a very frustrating experience because adequate doses of radiation cannot be delivered without significant risk to these patients. Several single-institution reports that used doses in the range of 20 to 30 Gy for the management of these patients are available in the literature, and in the majority of these studies, patients derived some clinical benefit but overall outcome remained poor.65,66 It is unclear whether reirradiation provided substantial benefit for these patients, and therefore when opting for reirradiation in such patients, significant attention must be paid to the individual patient’s history and clinical findings to determine whether there is any role for reirradiation. The decision of whether to repeat WBRT or use limited-field RT must also be made. Aggressive, primary treatment interventions to provide durable local control is obviously preferable to being backed into this unfortunate situation of recurrent intracranial disease with limited treatment options.
Primary Central Nervous Neoplasms
In 2008 there were an estimated 21,810 new cases of CNS malignancies in the United States and 13,070 deaths.67 The incidence of CNS and primary malignant brain tumors, excluding lymphoma, leukemia, pituitary, and olfactory tumors, is 6.4 cases per 100,000 person-years and is slightly higher in males.68 Meningiomas represent the only CNS tumor seen more commonly in females. Given the infiltrative nature of many CNS neoplasms and the essential and nonredundant nature of the majority of CNS structures restricting surgical excision, external beam RT has long held a role in the treatment of CNS malignancies.
Malignant Glioma
Malignant gliomas are relatively uncommon neoplasms that account for only about 40% of the approximately 21,810 new annual cases of CNS tumors in the United States in 2008, but they represent 80% of malignant CNS tumors.67 Their clinical impact continues to far outweigh their incidence, principally because of their high fatality rate. Malignant or high-grade gliomas are rapidly growing tumors that tend to diffusely infiltrate the brain parenchyma, thus typically preventing complete microscopic surgical excision. These tumors can develop at any age but are most commonly seen in late adulthood. Malignant gliomas are usually divided into grade III and grade IV gliomas, which have distinct molecular features, responses to treatment, and prognoses.69
Glioblastoma
Glioblastoma, or grade IV astrocytoma, accounts for about 80% of all malignant gliomas, and 1- and 5-year survival rates were only 29.3% and 3.3%, respectively, in an analysis of the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2002.68 Average survival has improved from 12 to 14 months with a combination of RT and adjuvant TMZ chemotherapy, but long-term survivorship remains low. Malignant gliomas such as glioblastoma rarely metastasize, and mortality is primarily attributed to nearly universal local failure, which leads to early institution of RT as an adjunct to surgery in an attempt to improve local control.
Utility of Radiation Therapy
Although retrospective analysis demonstrates that nearly gross total resection (>98%) of glioblastoma multiforme (GBM) leads to an improvement in overall survival,70 the infiltrative nature of GBM prohibits complete surgical eradication of disease without unacceptable neurological compromise. This limitation has led to the establishment of RT in the adjuvant setting as a cornerstone of treatment. An institutional review by Bouchard and coauthors in the late 1960s of patients given higher doses of radiation (50 to 60 Gy) found a survival rate of 44% at 1 year in those treated with a combination of surgical resection and adjuvant RT,71 which compared favorably with other institutions in which RT was not used. Intra-institutional comparisons in Sweden72 and at the University of California at San Francisco,73 University of California at Los Angeles,74 and the Mayo Clinic75 retrospectively identified improved survival when adjuvant RT was administered.
These favorable institutional data led to trials examining the adjuvant use of RT after surgical resection by the Brain Tumor Study Group (BTSG; later renamed the Brain Tumor Cooperative Groups [BTCG]), although the trials had patients with “malignant gliomas,” including anaplastic astrocytomas. The first randomized trial (BTSG 66-01) was actually designed to compare the activity of mithramycin postoperatively in the experimental arm versus surgery alone (Table 250-5).76–79 Although no difference based on the administration of mithramycin was identified, 55% of the patients received adjuvant RT at the discretion of the treating physician. Median survival was 36 weeks for those who received any dose of radiation and 15 weeks for those without. Patients receiving 50 Gy or greater had a median survival of 38.5 weeks; if less than 50 Gy was delivered, overall survival appeared to have been 33 weeks.
