SEARCH

SEARCH BY CITATION

Keywords:

  • glioblastoma;
  • radiation therapy utilization;
  • SEER;
  • surgical resection

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND:

The purpose of this study was to assess what factors influence radiation therapy (RT) utilization in patients with glioblastoma and to ascertain how patterns of care have changed over time.

METHODS:

A total of 9103 patients with supratentorial glioblastoma in the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2006 were analyzed. Demographic information was obtained, including age, sex, race, year of diagnosis, and marital status. Treatment characteristics included receipt of RT and surgical resection.

RESULTS:

In total, 76.8% of patients received RT, whereas 78% received resection. Patients of male sex, who were currently married, who were <65 years old, and who underwent resection were more likely to receive RT. The average annual percentage change in RT utilization in the years 1990-2006 was −0.41% (95% confidence interval [CI], −0.23 to −0.58), whereas for resection it was 0.26% (95% CI, 0.03 to 0.50). This equates to a 6.5% decrease in RT utilization and a 4.2% increase in resection during this time period. Patients treated with RT had a 2-year overall survival of 11.4%, compared with 5.2% in those not treated with RT (P < .00001). Multivariate analysis showed that younger age (continuous; odds ratio [OR], 0.97; P < .0001), marital status (OR, 1.62; P < .0001), surgical resection (OR, 1.72; P < .0001), and year of diagnosis 1998-2006 compared with 1990-1997 (OR, 0.82; P < .0001) were associated with RT utilization, whereas sex, lesion size, and race were not.

CONCLUSIONS:

SEER data show a decreasing utilization of RT in patients with glioblastoma from 1990 to 2006. Patients who were older, who were unmarried, and who underwent biopsy only were less likely to receive RT. Cancer 2012. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Glioblastoma is the most common primary brain malignancy in adults and continues to be associated with a very poor prognosis.1 Population-based data have shown an increasing incidence of glioblastoma, particularly in the elderly.1 Prospective randomized controlled trials have shown a doubling of survival for postoperative radiation therapy (RT) compared with best supportive care.2-4 More recently, temozolomide administered in conjunction with RT has been shown to be an effective systemic treatment that leads to improved survival among these patients.5 Thus, the current standard of care among patients with glioblastoma is maximal safe surgical resection followed by RT together with systemic concurrent and adjuvant chemotherapy with temozolomide.

Population-based studies using the Surveillance, Epidemiology, and End Results (SEER) database have shown that unmarried patients with glioblastoma are less likely to receive RT and resection and have worse survival than married patients.6 In addition, there exist racial differences in survival that may be explained by disparity in treatment.7, 8

The purpose of this study is to use a population-based database to assess what factors influence RT utilization in the United States and how patterns of care have changed over time.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Data Source

The SEER program of the National Cancer Institute assembles information on cancer incidence and survival in the United States. The SEER program registries routinely collect data on patient demographics, primary tumor site, tumor morphology and stage at diagnosis, first course of treatment, and follow-up for vital status. The registries participating in the SEER program capture approximately 97% of incident cases.9 The public use data contain information on type of surgery performed and whether a subject received RT. However, the registry does not contain information on comorbid conditions, imaging performed during staging evaluation, surgical margins, treatment details (such as RT dose and fields or systemic agents administered), or locoregional control. The population residing within the areas served by the SEER cancer registries is comparable to the general US population with regard to measures of poverty and education, but tends to be more urban and has a higher proportion of foreign-born persons than the general US population.10 The catchments for the 17 SEER registries used in this analysis comprise approximately 26% of the US population.11 As this data set is in the public domain, it was deemed institutional review board exempt.

Description of Study Cohort and Treatment

A total of 10,595 patients who were diagnosed with supratentorial World Health Organization grade IV glioblastoma (International Classification of Diseases for Oncology 3rd edition morphology codes 9440-9442) between January 1, 1973 and December 31, 2006 were identified in the SEER database using SEER*Stat software (version 7.0.4).12 Patients younger than 21 years (n = 111), who were not actively followed (diagnosed with death certificates or autopsy results only; n = 21), or who died within 1 month of diagnosis (n = 1360) were excluded. Patients who died within 1 month of diagnosis were excluded to reduce selection bias, as poor performance status and significant comorbidities may preclude these patients from receiving adequate therapy.13 Among these patients, 473 patients (34.8%) received biopsy only, and 887 patients (65.2%) underwent resection. Patient characteristics identified were age, race, year of diagnosis, and marital status. Currently married status, as defined by the SEER database, includes common-law marriages.

Treatment Course

Patients who received external beam RT and surgery were identified based on the SEER variables. Receipt of RT was defined as beam radiation (SEER codes “Beam radiation” or “Combination of beam with implants or isotopes”). Beam irradiation includes stereotactic radiosurgery. The extent of surgical resection was defined as biopsy only (1988-1997 SEER codes 1, 2, 3, 5, and 10; 1998-2006 SEER code 0) or resection (1988-1997 SEER codes 20, 30, 40, 50, 55, 60, 80, and 90; 1998-2006 SEER codes 10, 20, 40, 55, and 90). Given ambiguity in the SEER database regarding surgical coding, no attempt was made to differentiate subtotal from gross total resection.

Outcomes

Survival time was calculated from the diagnosis date to the death date through December 31, 2006. Overall survival (OS) was defined as the time between diagnosis and death from any cause.

Statistical Analysis

Data analysis was performed using Stata/SE 10.0 statistical software (Stata Corporation, College Station, Tex). Pearson chi-square was used to assess measures of univariate association in frequency tables. Unadjusted associations between treatment groups and outcomes were compared using survival analysis and the Kaplan-Meier log-rank test. A P value ≤.05 was considered to be statistically significant. Statistical tests were based on a 2-sided significance level. Average annual percentage change in RT and surgery utilization was calculated using a least-squares linear regression approach.

The Cox proportional hazard model was used for both univariate and multivariate analyses to assess the effect of patient characteristics on the endpoint. The endpoint for these analyses was receipt of RT. All variables were assessed on a univariate basis, and factors with a significance of ≤.25 were assessed for multivariate analysis using backward elimination. The estimated odds ratio (OR) is reported.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Patient demographics and tumor characteristics of the 9103 eligible patients are outlined in Table 1. Median follow-up for uncensored patients was 12 months (range, 2-147 months).

Table 1. Demographics and Tumor Characteristics (n=9103)
CharacteristicNo.%
  1. Abbreviation: NOS, not otherwise specified.

Histology  
 Glioblastoma NOS886097.3
 Giant cell glioblastoma790.9
 Gliosarcoma1641.8
Age (median, 63 years)  
 <50 years169218.6
 50-59 years206822.7
 60-74 years366540.3
 ≥75 years167818.4
Sex  
 Male524957.7
 Female385442.3
Race  
 White838392.1
 Black3684.0
 Other3523.9
Marital status  
 Married614267.5
 Unmarried296132.5
Tumor size  
 <3 cm2622.9
 ≥3 cm133514.7
 Unknown750682.5
Tumor location  
 Frontal lobe219224.1
 Temporal lobe220424.2
 Parietal lobe165218.1
 Occipital lobe4034.4
 Cerebrum NOS265229.1
Vital status  
 Alive8429.2
 Deceased826190.8

Treatment Characteristics

A total of 6987 patients (76.8%) received RT as part of their treatment. Surgical resection was performed in 7104 patients (78%), biopsy in 1768 patients (19.4%), and unknown procedure in 231 patients (2.5%). Several factors were associated with increased rates of RT utilization, including patients of male sex, patients who were currently married, patients <65 years old, and patients who underwent surgical resection (Table 2).

Table 2. Radiation Usage 1990-2006
CharacteristicRTNo RTP
No.%No.%
  1. Abbreviation: RT, radiation therapy.

Race     
 White643376.7195023.3.372
 Black27574.79325.3 
Sex     
 Male411580.798219.3<.0001
 Female287271.7113428.3 
Marrital Status     
 Married491780.1122519.9<.0001
 Not married207069.989130.1 
Age     
 <65 years411182.885617.2<.0001
 ≥65 years287669.5126030.5 
Surgery    <.0001
 Biopsy only119267.457632.6 
 Resection567579.9142920.1
 Unknown12051.911148.1 

Patterns of Care Over Time

In the modern era, RT utilization has decreased, whereas the rate of surgical resection has increased. Specifically, the average annual percentage change in RT utilization in the years 1990-2006 was −0.41% (95% confidence interval [CI], −0.23 to −0.58; Fig. 1). This finding equates to a 6.5% decrease in the rate of RT utilization during this time period. When stratified by sex, race, age, and surgical resection, this pattern persisted (Fig. 2). Conversely, the average annual percentage change in resection utilization in the years 1990-2006 was 0.26% (95% CI, 0.03 to 0.50; Fig. 3). This finding equates to a 4.2% increase in the rate of surgical resection during this time period.

thumbnail image

Figure 1. Radiation therapy utilization is shown among patients with glioblastoma from 1990 to 2006. CI, confidence interval.

Download figure to PowerPoint

thumbnail image

Figure 2. Average annual percentage change in radiation therapy utilization is shown from 1990 to 2006. CI, confidence interval.

Download figure to PowerPoint

thumbnail image

Figure 3. Surgical resection utilization is shown among patients with glioblastoma from 1990 to 2006.

Download figure to PowerPoint

OS

The median OS for those who received biopsy was 3 months (95% CI, 3-3), for resection it was 5 months (95% CI, 4-5), for biopsy with RT it was 5 months (95% CI, 5-6), and for resection with RT it was 10 months (95% CI, 10-11; Fig. 4). The 2-year OS for those who received biopsy was 2.5%, for resection it was 6.2%, for biopsy with RT it was 4.4%, and for resection with RT it was 12.9%. Patients treated with RT had a 2-year overall survival of 11.4%, compared with 5.2% in those not treated with radiation (P < .00001; Fig. 5).

thumbnail image

Figure 4. Overall survival is shown among patients with glioblastoma treated with biopsy only, resection only, radiation therapy (RT) only, or resection and RT.

Download figure to PowerPoint

thumbnail image

Figure 5. Overall survival is shown among patients with glioblastoma treated with or without radiation therapy.

Download figure to PowerPoint

Multivariate Analysis

Results of univariate and multivariate analyses are shown in Table 3. Multivariate analysis showed that younger age (continuous; OR, 0.97; P < 0001), married status (OR, 1.62; P < 0001), surgical resection (OR, 1.72; P < 0001), and year of diagnosis 1998-2006 compared with 1990-1997 (OR, 0.82, P < 0001) were associated with RT utilization, whereas sex, size of lesion, and race were not.

Table 3. Results of Univariate/Multivariate Analysis With Receiving Radiation Therapy as Endpoint
VariableUnivariate PMultivariate
RRP95% CI
  1. Abbreviations: CI = confidence interval; RR = relative ratio.

Final model    
 Age (continuous)<.00010.97<.00010.96-0.97
 Marital status (married vs unmarried)<.00011.62<.00011.46-1.80
 Surgery (surgery vs biopsy)<.00011.72<.00011.56-1.89
 Year of diagnosis (1998-2006 vs 1990-1997).0010.82<.00010.74-0.94
Full model    
 Male vs female<.0001   
 Size (continuous).60   
 Race (black vs white).37   

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

This study is the first published work of which we are aware that assesses the rate of RT utilization in glioblastoma over time in a population-based cohort. We found that patients who were older, who were unmarried, and who underwent biopsy only were less likely to receive RT. We did not find a racial difference in RT delivery, although there was only a small subset of black patients. In addition, our data suggest that RT utilization is decreasing over time, whereas surgical resection is increasing.

Multiple studies over the past 40 years have shown a benefit to RT in patients with glioblastoma, including elderly patients ≥70 years old.2-4 Thus, it is unclear why the utilization of RT in our study data set demonstrates a decrease during the time period of this study. In the past few years, some patients with glioblastoma may have been treated with temozolomide only, without RT. However, the results of the landmark trial by Stupp et al were not published until 2005; thus, the widespread use of temozolomide would not have been expected until after the time period examined by our investigation.5 Newer technologies developed during the time period of this study include 3-dimensional computed tomography-based treatment planning and intensity modulated RT, which reduce uncertainty about adequate dose delivery and spare normal structures. These technologies should have led to an increase in RT delivery, making these results quite surprising.

One possible explanation for the decrease in RT utilization is the increasing number of free-standing RT centers over this time period. Data from these centers may have not been accurately included in the SEER registry, making it appear that there is less use of RT over time. SEER data linked to Medicare claims found high agreement of the RT information in elderly patients, with 88% to 95% accuracy depending on disease site.14 Furthermore, independent verification of the RT record collection is needed to explore this potential cofounder.

There are several factors that may explain the increasing rate of surgical resection over time in our cohort. The increasing rate of surgical resection could be explained by improved imaging techniques, including magnetic resonance imaging, that allow better surgical planning. A second reason may be improvement in neurosurgical training in general. In addition, there is evidence that more extensive surgical resection is associated with improved progression-free survival.15 Given that the rate of RT is declining, we do not believe that the increase in utilization of surgical resection could be explained by a widespread, more aggressive treatment strategy in these patients.

The lower rate of RT utilization among those who receive biopsy only may be explained by a general lack of enthusiasm for multimodality treatment of glioblastoma multiforme (GBM) among community practitioners. This tendency to withhold treatment could be magnified in elderly patients based on the Radiation Therapy Oncology Group recursive partition analysis done in 1993 showing worse outcomes in older patients.16 Level I evidence emerged after the time frame of this study showing a benefit of RT in elderly patients with GBM, which could have lead to a change in the pattern of care subsequently.2

Although the current results are persuasive, several obstacles exist in addressing these questions using the SEER dataset alone. A recent study of SEER patients in Los Angeles and Detroit with nonmetastatic breast cancer showed a significant under-reporting of postoperative RT.17 About 12.5% of patients reported receiving RT, whereas their SEER record did not report the receipt of RT. Many of these patients may have received RT in a delayed fashion after neoadjuvant chemotherapy, and thus it was not recorded as a part of initial treatment. It is not clear how these results can be generalized to the entire SEER registry across other regions and disease sites. Given the current treatment strategy for glioblastoma, which includes upfront RT, this would be better recorded in our cohort. We find it unlikely that the decreasing rate of RT utilization could be solely explained by increasing rates of under-reporting in our cohort. For the current study, we did not examine how distance to treatment center influenced receipt of RT, although this has been shown to influence other disease sites.18-20 We also acknowledge that some outcomes in this study, particularly OS, are heavily influenced by confounding factors not specified in the SEER registry, including performance status, details of RT, chemotherapy usage, and extent of surgical resection. These and other factors will influence any study using the SEER database, and thus limit further interpretation of this data set.

Several studies have provided interesting insight into disparities in treatment and survival among patients with glioblastoma. Chang et al found that unmarried patients with glioblastoma presented with larger tumors, were less likely to receive surgical resection and RT, and had worse survival than married patients.6 In addition, SEER-Medicare data from patients ≥65 years old have shown that unmarried patients with more comorbidities were less likely to receive RT or chemotherapy.21 Finally, there exist racial differences in survival that may be explained by disparity in treatment,7, 8 although this finding was not noted in the current analyses. Depression is present in 6% to 28% of patients with GBM and could influence decisions regarding treatment, although this cannot be evaluated with this database.22, 23

In conclusion, despite widely accepted data from prospective randomized-controlled trials showing a significant survival benefit to RT, practice patterns in the United States show a decrease in utilization of RT in patients with glioblastoma over the past 16 years; in contrast, the rate of surgical resection has increased over this time. Further research is needed to determine whether this trend continues in the modern era and to ascertain physician biases that may limit the receipt of RT in patients with glioblastoma.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES