Presented in part at the American Society for Radiation Oncology 52nd Annual Meeting; October 31-November 4, 2010; San Diego, CA.
This study used the linked Surveillance, Epidemiology, and End Results (SEER) database. The authors acknowledge the SEER Program (www.seer.cancer.gov) and the National Cancer Institute in the creation of the SEER*Stat Database.
The interpretation and reporting of these data are the sole responsibility of the authors.
The role of postoperative radiotherapy (PORT) in the management of low-grade glioma remains controversial. An analysis using data from the European Organization for Research and Treatment of Cancer 22844/22845 studies concluded that several factors portend a poor prognosis: age ≥40 years, astrocytoma histology, tumor size ≥6 cm, tumor crossing midline, and preoperative neurologic deficits. PORT may benefit patients with high-risk features. The aim of this study was to assess temporal trends and determinants of the use of PORT.
By using data from the Surveillance, Epidemiology, and End Results program, the authors identified 1127 adult patients diagnosed with low-grade glioma (World Health Organization grade I and II) who underwent surgical resection between January 1, 1998 and December 31, 2006. The primary outcome was receipt of PORT. The authors performed multivariate logistic regression to examine the association between clinical, patient, and demographic characteristics and receipt of PORT.
Receipt of PORT declined during the study period, from 64% of patients in 1998 to 36% of patients in 2006. On multivariate analysis, significant predictors of receipt of PORT were age ≥40 years, tumor crossing midline, and partial surgical resection.
Low-grade glioma (LGG), defined by the World Health Organization (WHO) as grade I or II oligodendroglioma, astrocytoma, or mixed oligoastrocytoma, accounts for about 10% of all primary central nervous system tumors and 25% of gliomas.1 Although most LGGs are relatively slow-growing, they can behave heterogeneously, and therefore outcomes vary widely. Surgical resection is the primary modality of treatment, and the role of adjuvant radiation therapy remains controversial.
Two seminal randomized trials from the European Organization for Research and Treatment of Cancer (EORTC) have provided evidence to guide the postoperative management of low-grade glioma. EORTC 22844 randomized patients to 45 gray (Gy) versus 59.4 Gy for adjuvant postoperative radiation therapy (PORT), and showed no difference in progression-free survival (PFS) or overall survival (OS).2 EORTC 22845 randomized 314 patients to adjuvant PORT versus salvage radiation at the time of progression.3 At 8-year follow-up, adjuvant PORT improved PFS, but did not impact OS. An analysis of the EORTC 22844/22845 data, published by Pignatti et al in 2002, described 5 risk factors associated with lower overall survival: age ≥40 years, astrocytoma histology, tumor size ≥6 cm, midline extension, and preoperative neurologic deficits.4 The analysis showed that patients with ≥3 risk factors had nearly half the median OS of patients with fewer risk factors (3.7 years vs 7.8 years).
In the absence of robust clinical guidelines, clinicians have had to make individualized decisions about offering adjuvant PORT to patients with LGG status post resection. To date, no study has examined how adjuvant PORT has been incorporated into routine clinical practice outside of clinical trials. The objective of the current study was to investigate characteristics associated with receipt of adjuvant PORT for LGG and to assess temporal trends in its utilization.
MATERIALS AND METHODS
Data Source and Cohort
The study cohort consisted of cases of LGG reported to the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program from January 1, 1998 to December 31, 2006. The SEER registries collect incident cases of cancer in 6 metropolitan areas (Seattle, Detroit, San Francisco/Oakland, San Jose/Monterrey, Los Angeles, and Atlanta) and 11 states (Connecticut, New Jersey, Iowa, New Mexico, Utah, other California, Hawaii, Alaska, rural Georgia, Louisiana, Kentucky), accounting for approximately 26% of the US population.5
As shown in Figure 1, we selected 2384 patients with pathologic diagnosis of grade I or II malignant astrocytoma, oligodendroglioma, or mixed histology tumors between 1998 and 2006 (International Classification of Diseases for Oncology Third Edition Histology codes 9400/3, 9450/3, and 9382/3). Of those, 2116 were adults, defined as age older than 20 years, and 1957 had tumors located in the cerebrum, frontal lobe, temporal lobe, parietal lobe, occipital lobe, cerebellum, ventricle, and overlapping lesion of the brain (Site and Morphology codes C71.0-C71.6, C71.8). We then excluded patients with a history of prior malignancy (n = 131), patients who had no or unspecified surgery (n = 513), and patients with WHO grade I tumors with complete resection (n = 103), as they were excluded from participation in the EORTC trials. Patients who received preoperative radiation, brachytherapy, or intraoperative radiation (n = 83) were also excluded from our cohort, as these treatment modalities were not offered in the EORTC trials. Our final cohort consisted of 1127 patients, of whom 656 had surgery alone and 471 had surgery plus adjuvant PORT.
The primary outcome was receipt of adjuvant PORT. PORT was ascertained using the SEER treatment variable “beam radiation.” SEER treatment variables record radiotherapy if initiated within 4 months of diagnosis. Therefore, the primary outcome is most consistent with upfront or adjuvant PORT. In the EORTC trials, adjuvant radiation was defined as a maximum 8-week interval between surgery and initiation of radiation, with most patients initiating radiation after 4 weeks.
We were specifically interested in the risk factors identified in the Pignatti analysis: age (<40 or ≥40 years), histology (oligodendroglioma/mixed or astrocytoma), tumor size (<6 cm, ≥6 cm, or size unknown), and midline extension (does not cross midline, crosses midline, or extension unknown). Preoperative neurologic deficits were also identified by the Pignatti analysis but are not recorded in the SEER database and therefore could not be assessed. We also included other important clinical, patient, and demographic control variables. Year of diagnosis captured the combined effects of several factors that were changing over the study's time period, including publication of randomized evidence about the benefits of PORT, evolution in radiotherapy technique from conformal radiotherapy to intensity-modulated radiotherapy, and changes in physician reimbursement for treatment. Additional clinical, patient, and demographic control variables included primary site (supratentorial or infratentorial), extent of surgical resection (biopsy alone, partial resection, or gross total resection), sex (male or female), race (white, black, or other), marital status (unmarried, married, or marital status unknown), geography (Northeast, Midwest, West, or South), median income (≤$20,000, ≤$40,000, ≤$60,000, or >$60,000), and population (>1 million, 250,000 to 1 million, <250,000, or unknown).
We characterized temporal trends in care patterns in the time period before and after the publication of the Pignatti analysis. We assessed how the receipt of adjuvant PORT changed over the study time period across risk groups defined by the number of Pignatti risk factors.
We calculated the proportion of patients who received adjuvant PORT by dividing the number of patients who received PORT (numerator) by the total number of patients receiving surgery (denominator). We performed univariate and multivariate logistic regression to evaluate the association between receipt of adjuvant PORT, and clinical, patient, and demographic variables, with particular attention to the Pignatti risk factors. We report temporal trends in receipt of adjuvant PORT for the total sample of patients, as well as trends stratified by number of Pignatti risk factors. Statistical significance was set at .05, and all tests were 2-tailed. Statistical analysis was performed using JMP version 8.0 (SAS, Cary, NC). The study was approved by the institutional review board of the University of Pennsylvania.
Baseline Characteristics of Study Cohort
Of the 1127 patients in the cohort, 79% had supratentorial tumors, as compared with 21% with ventricular and cerebellar location. Oligodendroglioma or mixed histology represented 60% of patients, whereas 40% had astrocytoma histology. No tumor size was recorded in 35% of patients, and of the remaining patients, the majority had tumor size <6 cm. Greater than 80% of tumors did not cross the midline. Whereas the majority of patients had either a gross total or partial resection, 24% had biopsy alone. Forty-six percent of the population were younger than 40 years, 37% were female, and 82% were white. Most patients were married (55%), from the Western United States (54%), earned a median income of $40,000 or less (57%), and lived in a city with a population >1,000,000 (59%).
Table 1 presents clinical, patient, and demographic characteristics of the study cohort with reference to the EORTC cohorts used in the Pignatti analysis. Of note, the EORTC cohorts had a majority of astrocytoma histology, in contrast to the SEER cohort, which had majority oligodendroglioma/mixed histology. Although a large percentage of patients had unknown tumor size in SEER, of those with known tumor size, the proportion of <6 cm as compared with ≥6 cm was comparable to that of the EORTC cohort. A greater percentage of patients in EORTC 22844 had tumor crossing the midline as compared with EORTC 22845 and the SEER cohort. A greater percentage of patients had biopsy alone in the EORTC trials as compared with the SEER cohort, and the EORTC 22844 patients had a lower percentage of gross total resection as compared with EORTC 22845 and SEER. Age was distributed equally in all 3 cohorts.
Table 1. Study Cohort Characteristics as Compared With EORTC Trial
SEER Cohort, Total No. (%)
EORTC 22844, Total No. (%)
EORTC 22845, Total No. (%)
Abbreviations: EORTC, European Organization for Research and Treatment of Cancer; SEER, Surveillance, Epidemiology, and End Results.
Pignatti risk factors
Does not cross midline
Gross total resection
Receipt of PORT
Clinical, patient, and demographic characteristics of patients associated with receipt of adjuvant PORT are shown in Table 2. Of the 1127 patients in the study cohort, 471 (42%) received adjuvant PORT, and 656 (58%) did not. Of the Pignatti risk factors, only age (48% ≥40 years vs 36% <40 years; adjusted odds ratio [OR], 1.64; 95% confidence interval [CI], 1.25-2.16) and midline extension (68% crossing midline vs 40% without midline extension; adjusted OR, 2.85; 95% CI, 1.74-4.78) were significantly associated with PORT on multivariate analysis. In addition to age and midline extension, extent of resection was associated with PORT (57% with partial resection vs 43% with biopsy alone; adjusted OR, 1.71; 95% CI, 1.21-2.42). Those with gross total resection were significantly less likely to have adjuvant PORT compared with patients who underwent biopsy alone (adjusted OR, 0.50; 95% CI, 0.35-0.71). Large tumor size (adjusted OR, 1.05; 95% CI, 0.72-1.55) and astrocytoma histology (adjusted OR, 1.24; 95% CI, 0.94-1.65) were not significantly associated with PORT.
Table 2. Unadjusted and Adjusted Odds of Receiving PORT
Several other clinical and nonclinical factors were also associated with adjuvant PORT on adjusted analysis. Fewer patients who were unmarried received adjuvant PORT (46% married vs 37% unmarried; adjusted OR, 1.50; 95% CI, 1.13-2.00). In addition, there was substantial variation in receipt of adjuvant PORT by geographic region, ranging from 31% in Northeastern registries to 48% in Southern registries. Receipt of adjuvant PORT was not significantly associated sex, race, median income, or area population.
Overall, receipt of adjuvant PORT declined during the study period from 64% in 1998 to 36% in 2006 (Fig. 2). We grouped patients into risk categories based on risk features from the Pignatti analysis (age ≥40 years, astrocytoma histology, tumor ≥6 cm, or midline extension). Figure 2 demonstrates that in any given year, a higher percentage of patients with more poor prognostic features (ie, higher risk patients) received adjuvant PORT, although adjuvant PORT use declined in the study period for all groups.
We undertook this study to assess the use of adjuvant PORT for patients with low-grade glioma in a population-based cohort. We found that adjuvant PORT decreased across the study period in both high- and low-risk groups, but that patients with more high-risk features were more likely to receive PORT at all time points in the study period. We also found that some, but not all, risk factors identified in randomized clinical trials were significant predictors of adjuvant PORT in the SEER cohort. These findings suggest that features predicting for higher risk of progression in randomized clinical trials are not being interpreted as indications for PORT in real world clinical practice. The lack of overall survival benefit demonstrated in the EORTC data may have led to declines in PORT for all patients, irrespective of presence of risk factors. At present, no consensus guidelines exist to help clinicians decide which clinical features are indications for upfront PORT.
Limited translation of large clinical trial findings into real world clinical practice may be because of differences between patients treated in typical clinical settings and those treated in the 2 seminal EORTC studies. The most notable difference is histology; the majority of patients in EORTC 22844 and 22845 had astrocytoma histology, whereas the SEER cohort had majority oligodendroglioma/mixed histology. Oligodendroglioma originates from a distinct cell line as compared with astrocytoma, and prognosis is better because of its less aggressive nature, younger age at presentation, and higher sensitivity to chemotherapy.6 In addition, combined deletion of chromosomes 1p and 19q confer an OS and PFS advantage for oligodendroglioma histology.7 Mixed histology tumors are not as favorable as pure oligodendrogliomas, but are less aggressive than pure astrocytoma.6 For other variables, such as extent of surgical resection and midline tumor crossing, the EORTC cohorts differ from each other substantially. With regard to tumor size, age, and sex, the cohorts were equally distributed.
The differences between patients treated in typical care settings versus those treated in the EORTC trials raise concerns about the generalizability of the EORTC findings to the larger community of patients. When treating physicians look to the literature for guidance on postoperative management of low-grade glioma, they may not see data on patients that represent those seen in their practice. A lack of consensus guidelines that reflect the patient populations that are commonly seen in the community may lead to confusion over which patients need adjuvant PORT. As a result, some patients for whom adjuvant PORT is indicated may not be offered treatment at all.
In addition to EORTC 22844 and 22845, 2 other randomized trials and several retrospective studies have attempted to define prognostic features to stratify patients into high- and low-risk groups. Table 3 compares these studies; in addition to the Pignatti risk features, extent of surgical resection, tumor location, tumor grade, radiographic contrast enhancement, performance status, and molecular markers have all been found to be important prognostic factors in various retrospective studies.8-20 In the SEER cohort, we found that Pignatti risk factors of age ≥40 years and midline extension were associated with adjuvant PORT, whereas histology and tumor size were not. Extent of surgery was significant in our cohort and in several other studies, but not in the Pignatti risk features. Sex, race, and median income were not significantly associated with receipt of adjuvant PORT, despite the finding that nonclinical factors have been shown to affect receipt of postoperative radiation in other sites of disease.21-24 Clearly, there is no consensus about which factors truly differentiate a high-risk patient from a low-risk patient. Perhaps, in the absence of clear guidelines regarding indications for adjuvant PORT, its use has declined even for the patients most likely to benefit.
Table 3. Summary of Studies Discussing Guidelines for Adjuvant Postoperative Radiation Therapy in Low-Grade Glioma
Risk Features (MV Analysis)
Abbreviations: CSS, cause-specific survival; CT, computed tomography; EORTC, European Organization for Research and Treatment of Cancer; KPS, Karnofsky performance score; MV, multivariate; OS, overall survival; PFS, progression-free survival; post-op, postoperative; pre-op, preoperative; RCT, radiochemotherapy; RT, radiotherapy; SWOG, Southwest Oncology Group; UV, univariate.
1p19q status (age, tumor size, tumor crossing midline, contrast enhancement, extent of surgery, and seizures at diagnosis were significant on UV but not MV analysis)
It is also possible that neurosurgeons may not refer patients for radiation therapy or radiation oncologists may choose to defer treatment because of the lack of OS benefit demonstrated by both EORTC trials and the risk of side effects from therapy. Radiation therapy for LGG has been associated with impaired cognitive and executive function25 and potential fatal radiation necrosis even at low doses.8 In addition, patients in whom radiation therapy was delayed until the time of progression had high rates of salvage, and therefore the same OS as comparable patients treated upfront with PORT. Hence, without an obvious survival benefit or evidence to suggest that upfront radiation prolongs a higher quality of life, some neurosurgeons and radiation oncologists may possess inherent and reasonable bias against upfront radiation therapy. They may instead refer and treat patients only upon documented progression, even if a patient has multiple risk factors at the time of presentation.
Defining which subsets of patients will benefit from adjuvant PORT is challenging. More pragmatic randomized controlled trials are needed to ensure the patient population is representative of patients in the community. The term pragmatic has been used to describe trials that are designed to study interventions in the context in which they will be used, as opposed to ideal conditions. Incorporation of pragmatic principles can be achieved by more careful selection of exclusion criteria, identification of barriers of entry for diverse populations, and comparison to database records such as SEER.26-28 For example, future LGG clinical trials may be designed to enroll more patients with oligodendroglioma or mixed histology, more accurately reflecting patients seen in routine clinical practice.
There are several limitations of this study. First, given the 4-month interval of inclusion used for PORT in the SEER database, our study describes radiotherapy practice patterns most consistent with adjuvant PORT. We are unable to describe trends in the use of salvage radiotherapy in this cohort, and it is possible that adjuvant PORT has fallen in favor of salvage approaches. Second, 35% of patients did not have tumor size recorded. We note that 68% of patients with missing tumor size had biopsy alone or partial resection, which may have contributed to incomplete information about tumor size. Third, the impact of presurgical neurologic deficit, performance status (Eastern Cooperative Oncology Group >2 excluded by EORTC), comorbid illness (excluded by EORTC), and 1p19q deletion status cannot be assessed with the current database, and these factors may have affected candidacy for PORT. 1p19q deletion has been shown to impact outcomes and is now routinely used as a factor influencing the recommendation of adjuvant PORT.7
In conclusion, adjuvant PORT for low-grade glioma has declined in all risk groups, even patients deemed to be high risk by the Pignatti analysis. The subset of patients with LGG most likely to benefit from adjuvant PORT has not been defined, and there is a lack of consensus about which patient- and tumor-specific factors confer a higher risk of progression. Furthermore, existing randomized clinical trial data may not accurately reflect real world patients, and therefore incorporation of guidelines may be suboptimal. In the future, pragmatic studies investigating subsets of patients likely to benefit from adjuvant PORT will be critical in the management of LGG.