Abstract
BACKGROUND
The influence of subtotal resection (STR) on neurocognitive function (NCF), quality of life, and symptom burden in glioblastoma is unknown. If bevacizumab preferentially benefits patients with STR is unknown.
OBJECTIVE
To examine these uncertainties.
METHODS
NCF and patient reported outcomes (PRO) were prospectively collected in NRG Oncology RTOG 0525 and 0825. Changes in NCF and PRO measures from baseline to prespecified times were examined by Wilcoxon test, and mixed effects longitudinal modeling, to assess differences between patients who received STR vs gross-total resection. Changes were also compared among STR patients on 0825 receiving placebo vs bevacizumab to assess for a preferential therapeutic effect. Overall survival between STR and gross-total resection patients was compared using the Kaplan–Meier method.
RESULTS
A total of 427 patients were eligible with STR present in 37%. At baseline, patients with STR had worse NCF, worse MD Anderson Symptom Inventory Brain Tumor Neurological Factor ratings (P = .004), and European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (P = .002). Longitudinal multivariate analysis associated STR with worse NCF (Hopkins Verbal Learning Test–Revised Delayed Recognition [P = .048], Trail Making Test Part A [P = .035], and Controlled Oral Word Association [P = .049]). One hundred eighty-three STR patients from 0825 were analyzed (89 bevacizumab, 94 placebo); bevacizumab failed to demonstrate improvement in select NCF or PRO measures.
CONCLUSION
STR patients had worse NCF and PROs before therapy. During adjuvant therapy, STR patients had worse objective NCF, despite accounting for tumor location. STR did not result in a detriment to OS. The addition of bevacizumab did not preferentially improve PRO or NCF outcomes in STR patients.
Keywords: Resection status and quality of life, Resection extent and neurocognitive function, Glioblastoma resection, Glioblastoma and extent of resection, Extent of resection in GBM, Quality of life GBM, patient reported outcomes GBM, Bevacizumab and GBM, Glioblastoma and radiation therapy, Glioblastoma and surgical resection
ABBREVIATIONS
- COWA
Controlled Oral Word Association
- EORTC-QLQ
European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire
- GBM
glioblastoma
- GTR
gross-total resection
- HVLT-R
Hopkins Verbal Learning Test–Revised
- KPS
Karnofsky Performance Status
- MDASI-BT
MD Anderson Symptom Inventory Brain Tumor
- MRI
magnetic resonance imaging
- NCF
neurocognitive function
- NIH
National Institutes of Health
- OS
overall survival
- PFS
progression-free survival
- PROs
patient reported outcomes
- QOL
quality of life
- STR
subtotal resection
- TMT
Trail Making Test
Glioblastoma (GBM) is a highly infiltrative and aggressive malignancy with overall survival (OS) rates that remain dismal.1,2 Attempts to improve OS in newly diagnosed GBM using dose dense schedules of chemotherapy or novel targeted agents including bevacizumab have largely failed.3,4 Exciting advances in surgical techniques including 5-ALA guided surgery and intraoperative magnetic resonance imaging (MRI)-assisted surgery have recently emerged; however, quantifying the precise benefits of these technologies is challenging.5 Several publications have described an extent of resection “threshold” that distinguishes patients with higher median OS.6-8 Yet others have used accelerated failure time modeling and have recommended a maximum safe resection approach.9 Notable is that more extensive resections may carry a higher risk of morbidity and neurological impairment. Additionally, in these patients with relatively short anticipated survival, an understanding of the influence of interventions on neurocognitive function (NCF) and patient reported outcomes (PROs), and quality of life (QOL) seems highly prudent. Nevertheless, only a limited number of small series have examined the relationship between resection status and such endpoints in patients with GBM.10-12 Changes in the normal brain function, specifically in NCF, are important in assessing the consequence of aggressive resection; further, NCF has an important role in GBM prognosis.13 However, there is a paucity of data examining the impact of subtotal resection (STR), relative to gross-total resection (GTR) on NCF in large cohorts of prospectively collected data. There are no published manuscripts, to the best of our knowledge, which have collectively examined NCF and PROs as influenced by the extent of resection.
NRG Oncology 0525 was a phase III randomized clinical trial comparing conventional adjuvant temozolomide chemotherapy with dose intensive temozolomide.4 NRG Oncology RTOG 0825 was phase III randomized clinical trial evaluating the use of bevacizumab in patients with newly diagnosed GBM.3 These clinical trials were 2 of the largest clinical trials in GBM ever conducted, and both were the first to collect detailed data examining NCF and PROs.14,15 These data provided the first opportunity to examine the relationship between the extent of resection and the NCF and PRO metrics over time.
Bevacizumab is a humanized monoclonal antibody against Vascular Endothelial Growth Factor A (VEGF-A) and preclinical data suggest that its therapeutic benefit is likely secondary to normalization of tumor vasculature.16,17 The clinical trial NRG Oncology RTOG 0825 demonstrated that the addition of bevacizumab to the upfront treatment of newly diagnosed GBM did not improve OS or NCF or PRO metrics.3,15 However, the possibility that STR patients may have intact tumor vasculature lead to the hypothesis that bevacizumab may preferentially improve clinical metrics in these patients.17 This analysis sought to simultaneously examine the influence of bevacizumab in only those patients with a STR.
METHODS
Sample
This was a posthoc exploratory analysis conducted using the merged dataset from the companion studies of NRG Oncology RTOG 0525 and NRG Oncology RTOG 0825 in which NCF and PRO data were prospectively collected. The merger of the 0525 and 0825 datasets was felt to be appropriate to add statistical power given extremely similar inclusion criteria of the 2 parent studies. The primary analysis for each subcomponent of the publication has been previously presented.14,15 Patients were included from only the control arm of 0825 to avoid the known influence and potential confounding factor of bevacizumab on NCF and PRO metrics15. All patients who participated in these subcomponents could speak English and had histopathologically confirmed GBM (World Health Organization grade IV). All patients provided written informed consent for participation in each of the clinical trials. Eligible patients with unknown resection status or biopsies only were removed from the analysis. Patients with biopsies only were removed secondary to their ineligibility for study participation. The distinction of GTR and STR was made by the accruing site and was based on postoperative computed tomography or MRI reports from the accruing site. The reason for STR was based on the discretion of the surgeon at the accruing center, and was likely secondary to tumor in more eloquent locations. Radiation therapy treatment specifications were the same in both clinical trials RTOG 0525 and RTOG 0825.
Study Design
The objective of this study was to compare changes in QOL, symptom burden, and NCF between 2 cohorts of patients, STR and GTR, in NRG Oncology RTOG 0525 and NRG Oncology RTOG 0825. Patients were only eligible for this portion of the clinical trial if they had not demonstrated radiographic progression during therapy. In addition, another objective of this study was to examine if the addition of bevacizumab preferentially improved QOL, symptom burden, and NCF metrics in patients with a STR. There were a total of 4 follow-up time points of interest for data collection: prior to cycles 1, 4, and 10, corresponding to 6, 21, and 42 wk from the start of treatment in NRG Oncology RTOG 0525 and at 6, 22, and 46 wk from the start of treatment in NRG Oncology RTOG 0825. Because of the 4 wk discrepancy in the last follow-up time point, these data were only incorporated into certain analyses in which an adjustment for the study variable (0525 vs 0825) could be included. Each patient's change from baseline to these follow-up time points (prior to cycles 1, 4 in NRG Oncology RTOG 0525, and 6 and 22 wk in NRG Oncology RTOG 0825) was calculated. Patients who had undergone STR were then selected from NRG Oncology RTOG 0825 to compare the placebo vs the bevacizumab arms.
Outcome Measures
The results of these companion clinical trials of NRG Oncology RTOG 0525 and RTOG 0825 have been previously reported.14,15 Health-related QOL was evaluated using the European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire C30/BN20 (EORTC-QLQ-C30/BN20) developed and validated for use in a cancer patient population.18,19 For this analysis, prespecified domains were global QOL and cognitive function from the EORTC-QLQ-C30 and communication deficit and motor dysfunction from the BN20. Symptom burden and interference were assessed using the MD Anderson Symptom Inventory Brain Tumor (MDASI-BT), which was developed and validated for use in this patient population.14,20 Symptom severity, cognitive factor, neurological factor, and interference were prespecified for this analysis.
NCF was assessed using the Hopkins Verbal Learning Test–Revised (HVLT-R; Total Recall, Delayed Recall, and Delayed Recognition),21 the Trail Making Test (TMT; Part A and Part B)22 and the Controlled Oral Word Association (COWA) test.23 The NCF tests were administered by a health care professional (eg, nurse, psychologist) who was trained and certified by the study neuropsychologist (J.S.W.). The Clinical Trial Battery Composite score, calculated using the standardized HVLT-R, TMT Parts A and B, and COWA scores, was also assessed.
Statistical Methods
After conducting a normality test, either the t-test or Wilcoxon test was used to assess resection status differences at these time points. Mixed effects models with random intercepts and slopes were used to assess changes across time. Adjustment was made for resection status (STR vs GTR), study (0525 vs 0825), and any baseline characteristics that were significantly different (at the α = 0.05 level) between type of resection status. Other covariates that were considered for inclusion were age (<60 vs ≥ 60 yr old), RPA (recursive partitioning analysis) class (III vs IV vs V), MGMT (O6-methylguanin-DNA-methyltransferase) status (methylated vs unmethylated), neurological function (no symptoms vs at least some symptoms), gender (male vs female), Karnofsky Performance Status scale (KPS; 60-80 vs 90-100), tumor laterality (right vs left), tumor location, temozolomide dosing (days 1-5 vs days 1-21), baseline anticonvulsant use (yes vs no), and baseline steroid use (yes vs no). Models were built as follows: covariates significant in a univariable model at the 0.10 level were considered for inclusion in the multivariable model. Covariates were removed one at a time, beginning with the covariate with the highest corresponding P-value, until all covariates remaining were significant at the 0.05 level. Higher alpha levels were used in the model building process due to its exploratory nature.
The effect of time from surgery to start of treatment, tumor laterality, and tumor location on baseline NCF and PRO metrics was examined using a linear regression model. The change from baseline to 2 of the follow-up time points (prior to cycles 1 and 4 in 0825) was calculated as follow-up―baseline. After conducting a normality test, either the t-test or Wilcoxon test were used to assess resection status differences at these time points. To adjust for the multiple domains being analyzed, while accounting for the correlated nature of these domains, a significance level of 0.01 was used.
OS and progression-free survival (PFS) for the cohort as a whole were estimated using the Kaplan–Meier method. Differences between the STR and GTR patients were assessed using the log rank test. A Cox proportional hazards model was used to calculate hazard ratios.
RESULTS
Patient Cohort
A total of 427 eligible patients who participated in the NCF and PRO components of NRG Oncology RTOG 0525 and the control arm of NRG Oncology RTOG 0825 with either an STR or GTR were included in the initial portion of the analysis (Figure 1A). Patients with unknown resection status or biopsies only were removed. A second cohort of patients were analyzed, those with STR on NRG Oncology RTOG 0825, this consisted of a total of 183 patients (94 patients in the control arm and 89 patients in the treatment arm; Figure 1B). Complete patient characteristics, with statistical comparisons between the different arms of the studies, are in Tables 1 and 2. There were more patients who underwent a GTR (n = 268) compared to STR (n = 159). Both groups were similar with respect to all patient and clinical characteristics except KPS. A total of 60.4% of patients with a STR had a KPS of 90 to 100 compared to 70.5% of patients with a GTR (Table 1; P = .032). The characteristics of the patients with a STR that were analyzed from NRG Oncology RTOG 0825 can be seen in Table 2. There were no statistically significant differences seen amongst the patient cohorts in this subgroup.
FIGURE 1.
A, CONSORT diagram of NRG Oncology RTOG 0525 and RTOG 0825 control arm patients. B, CONSORT diagram of NRG Oncology RTOG 0825 STR patients.
TABLE 1.
Baseline Patient Characteristics
| Partial resection (n = 159) | Total Resection (n = 268) | P-valuea | |
|---|---|---|---|
| Age (yr) | |||
| Mean | 57.3 | 56.2 | .37b |
| Std. dev. | 12.1 | 12.4 | |
| Median | 59 | 57 | |
| Min - max | 19 - 82 | 20 - 84 | |
| Q1 - Q3 | 50 - 66 | 50 - 64 | |
| Gender | |||
| Male | 98 (61.6%) | 150 (56.0%) | .25 |
| Female | 61 (38.4%) | 118 (44.0%) | |
| KPS | |||
| 60-80 | 63 (39.6%) | 79 (29.5%) | .032 |
| 90-100 | 96 (60.4%) | 189 (70.5%) | |
| Neurological function | No symptoms vs at least some | .053 | |
| No symptoms | 47 (29.6%) | 104 (38.8%) | |
| Minor symptoms | 72 (45.3%) | 118 (44.0%) | |
| Moderate symptoms | 40 (25.2%) | 45 (16.8%) | |
| Severe symptoms | 0 (0.0%) | 1 (0.4%) | |
| MGMT | Methylated vs unmethylated | .52 | |
| Methylated | 44 (27.7%) | 78 (29.1%) | |
| Unmethylated | 109 (68.6%) | 167 (62.3%) | |
| Unknown | 6 (3.8%) | 23 (8.6%) | |
| RPA class | |||
| III | 30 (18.9%) | 51 (19.0%) | .31 |
| IV | 97 (61.0%) | 178 (66.4%) | |
| V | 32 (20.1%) | 39 (14.6%) | |
| Lateralization of tumor | Right vs left | .22 | |
| Right side only | 80 (50.3%) | 155 (57.8%) | |
| Left side only | 74 (46.5%) | 112 (41.8%) | |
| Bilateral | 4 (2.5%) | 1 (0.4%) | |
| Unknown | 1 (0.6%) | 0 (0.0%) | |
| Tumor location | Single vs multiple | .18 | |
| Frontal | 43 (27.0%) | 78 (29.1%) | |
| Temporal | 44 (27.7%) | 75 (28.0%) | |
| Parietal | 24 (15.1%) | 44 (16.4%) | |
| Occipital | 0 (0.0%) | 5 (1.9%) | |
| Other | 0 (0.0%) | 1 (0.4%) | |
| Multiple | 48 (30.2%) | 65 (24.3%) | |
| Days from surgery to start of treatment | |||
| Mean | 28.6 | 28.6 | .91b |
| Std. dev. | 5.0 | 5.5 | |
| Median | 29 | 28 | |
| Min - max | 3 - 36 | 10 - 64 | |
| Q1 - Q3 | 25 - 33 | 25 - 33 | |
| Steroid use | |||
| No | 33 (20.8%) | 62 (23.1%) | .57 |
| Yes | 126 (79.2%) | 206 (76.9%) | |
| Anticonvulsant use | |||
| No | 46 (28.9%) | 64 (23.9%) | .25 |
| Yes | 113 (71.1%) | 204 (76.1%) | |
| TMZ dose | |||
| TMZ d 1-5 | 127 (79.9%) | 209 (78.0%) | .64 |
| TMZ d 1-21 | 32 (20.1%) | 59 (22.0%) | |
| Treatment arm | |||
| 0525 arm 1 | 33 (20.8%) | 63 (23.5%) | .64 |
| 0525 arm 2 | 32 (20.1%) | 59 (22.0%) | |
| 0825 arm 1 | 94 (59.1%) | 146 (54.5%) | |
| Study | |||
| 0525 | 65 (40.9%) | 122 (45.5%) | .35 |
| 0825 | 94 (59.1%) | 146 (54.5%) | |
TMZ, temozolomide; Std. dev., standard deviation; KPS, Karnofsky performance score; MGMT, O6-methylguanin-DNA-methyltransferase; Q1, first quartile; Q3, third quartile; RPA, recursive partitioning analysis
a P-value from Chi-Square test
b P-value from t-test assuming equal variances
TABLE 2.
Baseline Characteristics by Treatment Arm for 0825 STR Patients
| Control (n = 94) | Bev (n = 89) | P-valuea | |
|---|---|---|---|
| Age (yr) | |||
| Mean | 57.6 | 59.3 | .29b |
| Std. dev. | 11.7 | 10.4 | |
| Median | 58.5 | 60 | |
| Min - max | 19 - 82 | 29 - 82 | |
| Q1 - Q3 | 51 - 66 | 54 - 66 | |
| Gender | |||
| Male | 63 (67.0%) | 53 (59.6%) | .29 |
| Female | 31 (33.0%) | 36 (40.4%) | |
| KPS | |||
| 60-80 | 40 (42.6%) | 32 (36.0%) | .36 |
| 90-100 | 54 (57.4%) | 57 (64.0%) | |
| Neurological function | |||
| No symptoms | 26 (27.7%) | 21 (23.6%) | .53 |
| Minor symptoms | 41 (43.6%) | 48 (53.9%) | |
| Moderate symptoms | 27 (28.7%) | 20 (22.5%) | |
| MGMT | |||
| Methylated | 24 (25.5%) | 22 (24.7%) | .99 |
| Unmethylated | 70 (74.5%) | 64 (71.9%) | |
| Unknown (indeterminate, invalid) | 0 (0.0%) | 3 (3.4%) | |
| RPA class | |||
| III | 14 (14.9%) | 9 (10.1%) | .41 |
| IV | 58 (61.7%) | 63 (70.8%) | |
| V | 22 (23.4%) | 17 (19.1%) | |
| Lateralization of tumor | |||
| Right side only | 44 (46.8%) | 48 (53.9%) | .28 |
| Left side only | 46 (48.9%) | 36 (40.4%) | |
| Bilateral | 3 (3.2%) | 4 (4.5%) | |
| Unknown | 1 (1.1%) | 1 (1.1%) | |
| Tumor locationc | |||
| Frontal | 21 (22.3%) | 18 (20.5%) | .20 |
| Temporal | 23 (24.5%) | 15 (17.0%) | |
| Parietal | 16 (17.0%) | 11 (12.5%) | |
| Occipital | 0 (0.0%) | 2 (2.3%) | |
| Other | 0 (0.0%) | 2 (2.3%) | |
| Multiple | 34 (36.2%) | 40 (45.5%) | |
| Days from surgery to start of treatmentc | |||
| Mean | 29.1 | 29.2 | .92b |
| Std. dev. | 5.0 | 5.7 | |
| Median | 30 | 30 | |
| Min - max | 3 - 36 | 7 - 39 | |
| Q1 - Q3 | 26 - 33 | 26 - 34 | |
| Steroid use | |||
| No | 21 (22.3%) | 20 (22.5%) | .98 |
| Yes | 73 (77.7%) | 69 (77.5%) | |
| Anticonvulsant use | |||
| No | 25 (26.6%) | 21 (23.6%) | .64 |
| Yes | 69 (73.4%) | 68 (76.4%) |
TMZ, temozolomide; Std. dev., standard deviation; KPS, Karnofsky performance score; MGMT, O6-methylguanin-DNA-methyltransferase; Q1, first quartile; Q3, third quartile; RPA, recursive partitioning analysis
a P-value from Chi-Square test
b P-value from t-test assuming equal variances
cOne patient on the Bev arm is missing and not included in the test between arms
NCF and PRO Comparison
Comparing the baselines scores from the various metrics measured for patients with STR in comparison to those with GTR, STR patients were found to have worse NCF (HVLT-R Total Recall [median score –1.5 for STR vs –1.0 for GTR, P = .004], COWA [–1.1 vs –0.8 for STR and GTR, respectively, P = .007]); worse MDASI-BT Cognitive Factor ratings (1.0 vs 0.5 for STR and GTR, respectively, P = .006); and worse EORTC-QLQ Cognitive Function scale ratings (66.7 vs 83.3 for STR and GTR, respectively, P = .002) at baseline. Those baseline comparisons reaching statistical significance are summarized in Table 3. Of note there were multiple baseline comparisons for each domain that did not reach statistical significance that are listed in the footer of Table 3.
TABLE 3.
Significant Baseline Distribution Differences for Net Clinical Benefit Metrics
| Partial resection | Total resection | P- | |
|---|---|---|---|
| (n = 159) | (n = 268) | valuea | |
| NCF | |||
| HVLT-R Total Recall | (n = 148) | (n = 242) | |
| Mean | –1.7 | –1.2 | |
| Std. dev. | 1.7 | 1.5 | |
| Median | –1.5 | –1.0 | .0031 |
| Min - max | –6.4 - 1.6 | –6.2 - 2.0 | |
| Q1 - Q3 | –2.9 - –0.3 | –2.1 - 0.1 | |
| COWA | (n = 149) | (n = 245) | |
| Mean | –1.1 | –0.7 | |
| Std. dev. | 1.3 | 1.3 | |
| Median | –1.1 | –0.8 | .0055 |
| Min - max | –4.1 - 2.1 | –3.7 - 5.0 | |
| Q1 - Q3 | –1.9 - –0.4 | –1.5 - 0.1 | |
| MDASI-BT | |||
| Neurological factor | (n = 149) | (n = 250) | |
| Mean | 1.3 | 1.0 | |
| Std. dev. | 1.6 | 1.5 | |
| Median | 0.7 | 0.3 | .0039 |
| Min - max | 0.0 - 8.0 | 0.0 - 9.3 | |
| Q1 - Q3 | 0.0 - 1.7 | 0.0 - 1.3 | |
| EORTC QLQ C30/BN20 | |||
| Cognitive function | (n = 149) | (n = 249) | |
| Mean | 70.1 | 77.4 | |
| Std. dev. | 24.5 | 23.1 | |
| Median | 66.7 | 83.3 | .0015 |
| Min - max | 0.0 - 100.0 | 0.0 - 100.0 | |
| Q1 - Q3 | 66.7 - 83.3 | 66.7 - 100.0 | |
HVLT-R, Hopkins Verbal Learning Test–Revised; COWA, Controlled Oral Word Association; Q1, first quartile; Q3, third quartile; Std. dev., standard deviation
a P-value from 2-sided Wilcoxon test using the normal approximation
Nonstatistically significant NCF baseline differences examined included: HVLT-R Delayed Recall, HVLT-R Delayed Recognition, TMT Part A, TMT Part B.
Nonstatistically significant MDASI-BT baseline differences examined included: Symptoms Severity, Neurological Factor, and Interference scores.
Nonstatistically significant baseline differences in the EORTC QLQ C30/BN20 include: Global QOL, Communication Deficit, Motor Dysfunction
When the discrete time point analysis was performed examining patient changes from baseline to the 6-, and the 22-wk time points, there were no statistically significant differences between the patients achieving a GTR as compared to STR (Tables, Supplemental Digital Content 1-6). On longitudinal multivariate analysis, several factors emerged as significantly associated with STR. The presence of STR was associated with worse NCF (HVLT-R Delayed Recognition [estimate = –0.018, P = .048], TMT Part A [–0.046, P = .035], and COWA [–0.0086, P = .049]) in patients without progressive disease. These differences can be seen in Table 4 along with graphical representation showing longitudinal trends in Figure 2. These differences were independent of tumor location and laterality, as both variables were accounted for in the statistical models. Resection status was not significantly associated with QOL or symptom burden over time.
TABLE 4.
Multivariable Longitudinal Analysis
| Effect | Estimate (std dev) | P-value |
|---|---|---|
| HVLT-R Delayed Recognition Multivariable Model | ||
| Intercept | –0.22 (0.73) | .76 |
| Time | –0.066 (0.033) | .046 |
| Surgery (total resection) | –0.095 (0.20) | .63 |
| Time*surgery (total Resection) | 0.018 (0.0089) | .048 |
| Gender (male) | 0.48 (0.14) | .0004 |
| KPS (90-100) | –0.61 (0.19) | .0014 |
| Tumor laterality (right side) | –0.94(0.18) | <.0001 |
| Study (0825) | 0.30(0.18) | .092 |
| TMT Part A Multivariable Model | ||
| Intercept | –4.82(1.16) | <.0001 |
| Time | 0.052 (0.013) | <.0001 |
| Surgery (total resection) | 0.073(1.03) | .94 |
| Time*surgery (total resection) | –0.046 (0.022) | .035 |
| KPS (90-100) | –0.63 (0.17) | .0002 |
| RPA III (V) | 2.91 (1.10) | .58 |
| RPA IV (V) | 1.75 (0.92) | .0086 |
| Study (0825) | 0.37(0.53) | .47 |
| COWA Multivariable Model | ||
| Intercept | –0.91 (0.21) | <.0001 |
| Time | 0.0090 (0.0025) | .0003 |
| Surgery (total resection) | –0.36 (0.12) | .0034 |
| Time*surgery (total resection) | –0.0086 (0.0043) | .049 |
| Gender (male) | 0.28 (0.12) | .18 |
| KPS (90-100) | –0.40 (0.15) | .0062 |
| Tumor laterality (right side) | –0.63 (0.11) | <.0001 |
| RPA III (V) | 0.73 (0.23) | .0016 |
| RPA IV (V) | 0.45 (0.18) | .012 |
| Study (0825) | –0.031 (0.12) | .79 |
Study, surgery, interaction, and KPS were forced into the model.
Reference levels are in parentheses.
FIGURE 2.
Longitudinal changes in net clinical benefit metrics. A, HVLT-R Delayed Recognition. B, TMT Part A. C, COWA.
Bevacizumab Subgroup Analysis
Patients with a STR only in NRG Oncology RTOG 0825 (n = 183) were examined from both the control and bevacizumab arms (n = 94, 89, respectively). The median change score was analyzed from baseline to 22 wk, comparing the control arm to those patients in the treatment arm (bevacizumab). The objective of this analysis was to examine if the addition of bevacizumab had a preferential benefit for patients with STR. There were no differences in any of the selected metrics of NCF, EORTC QLQ, or MDASI-BT for those patients with STR who received bevacizumab as compared to placebo (Tables, Supplemental Digital Content 7-9).
OS and PFS Comparison
Finally, OS and PFS were examined for all patients, comparing STR to GTR. No statistically significant difference in OS was seen. A significant improvement in PFS was observed in the GTR cohort (Figure 3; hazard ratio of 0.76 [95% confidence interval: 0.61, 0.94], P = .010).
FIGURE 3.
A, PFS and B, OS comparison between STR and GTR.
DISCUSSION
Our understanding of the influence of therapeutic interventions in GBM is in need of considerable expansion. OS and PFS are critical endpoints; however, short of the recently published EF-14 trial, major clinical breakthroughs in GBM have been limited.24 Considering the dismal outcomes in GBM, there is a need for improved understanding of current interventional strategies on NCF and PROs. This enables evaluation of interventions beyond just their influence on OS and PFS. Such an understanding might enable a more refined evaluation of both existing, and innovative therapeutic interventions. The precise influence of surgical resection in GBM is moderately controversial; however, most evidence supports a maximum safe resection.5-7,9 Yet the relationship between the presence of residual disease following surgical resection on patient outcomes such as QOL, symptom burden, and NCF remains largely unknown in GBM. Moreover, the ability of bevacizumab to preferentially improve NCF and PROs in those patients with an STR is also unknown. In the current report, a combined, posthoc, exploratory analysis of NRG Oncology RTOG 0525 and NRG Oncology RTOG 0825 is presented with the objective of addressing these questions.
These findings demonstrate that residual disease following STR has important implications in GBM. At baseline, patients with STR had worse NCF, MDASI-BT and worse EORTC-QLQ scale ratings, compared to GTR. This is likely secondary to tumor in more eloquent locations. Such differences in baseline scores may have important consequences for interpretation of future studies examining these endpoints. Possibly more interesting are the results of the longitudinal multivariate analysis in which patients with STR demonstrated worse NCF. This finding supports the widely held dictum in neuro-oncology that “tumor in the brain is bad for cognition.” Critical to this result is that collection of these NCF and PRO metrics was suspended if radiographic progression was demonstrated. This means that such declines were measured, for this analysis, in patients without clinical or radiographic evidence of progression. This removes tumor progression as a variable that could have accounted for NCF decline. This finding expands on and agrees with previously published data suggesting that NCF worsening precedes radiographic progression.25 The inclusion of NCF tests may allow identification of patients at risk for tumor progression before MRI evidence of tumor progression develops or differential impact on cognition from subsequent therapy. Moreover, the ability to measure these declines provides compelling data for a potentially novel endpoint for clinical trials, especially as OS did not vary between the GTR and STR patients.
Several nonstatistically significant findings of the study are worthy of mention. No significant influence of the time from surgery to the start of chemo-radiation therapy and baseline NCF, PRO, or QOL metrics or longitudinal metrics was observed. This is despite the potential for rapid tumor growth as illustrated in previously published series.26 Upon the revelation that patients with STR appeared to have worse select NCF and PRO metrics longitudinally, the influence of bevacizumab on these metrics was examined only in those patients with STR. The hypothesis was that the addition of bevacizumab in patients with STR may result in preferential improvement in these metrics. This was suspected secondary to the mechanism of action of bevacizumab, which centers on normalization of tumor vasculature. To this end a small, hypothesis generating, exploratory analysis of only those patients with STR was conducted. In this cohort, the change from baseline to 22 wk between patients receiving bevacizumab as compared to those patients receiving placebo was examined. Unfortunately, no preferential benefit was seen in those patients receiving bevacizumab with STR without demonstration of radiographic progression. This was consistent with the findings of the cohort as a whole.15 Given the absence of any benefit to bevacizumab in this smaller sub-group, we elected not to continue conducting exploratory analyses with bevacizumab as additional hypothesis driven subgroup analyses could not be formulated.
The influence of STR on OS and PFS was also examined; however, this was not the primary objective of this study and should be interpreted with caution. This analysis showed no difference with regard to OS for those patients with STR vs GTR. Patients with STR did have inferior PFS. Given there was no volumetric analysis performed to examine for a “threshold effect” of the extent of surgical resection these OS data are difficult to interpret. These OS comparisons should be interpreted with caution as this was not the primary focus of the analysis and the lack of volumetric analysis makes any firm conclusions from these data limited. These data support a maximum safe resection approach in terms of superior PFS5,7,9.
Limitations
There are limitations of this analysis that merit discussion. This analysis included only those patients with a high enough performance status to enroll in the clinical trials NRG Oncology RTOG 0525 and 0825. This may not be an accurate reflection of a typical GBM population. The studies did not perform central imaging analysis and as mentioned previously volumetric analysis of residual tumor was unknown. The absence of central imaging analysis prevented any ability to analyze the influence of residual tumor volume on OS or examine radiation dose distribution and its relation to the included metrics. Finally this is a posthoc exploratory analysis, which has inherent limitations. While we attempted to statistically control for nearly all potential confounders, data on some clinical metrics, such as steroid dosing was limited and was not able to be retrospectively ascertained with accuracy.
Limitations to this data not withstanding there are some potentially novel and intriguing aspects to this analysis that may carry clinical implications. For the first time we have shown that NCF and PRO metrics change on longitudinal multivariate analysis amongst patients with STR compared to GTR. Given these results, patients with STR may need to be more closely monitored for cognitive decline. In addition, the presence of STR should be accounted for in future studies examining NCF and PRO metrics. Furthermore, given that NCF tests were more sensitive to the detection of this decline, when compared with PROs, inclusion of neuropsychologists in the care of these patients may be of clinical benefit. Despite hypothetical benefits, we are seeing no signal for a preferential benefit to bevacizumab use in STR patients.3,15 Finally our data demonstrate a PFS benefit associated with GTR supporting a maximum safe resection approach.
CONCLUSION
These analyses demonstrate that NCF and PRO metrics differ in those patients achieving a STR as compared with a GTR. This is despite the absence of radiographic progression or difference in OS. These finding warrant additional investigation and validation in independent cohorts. These data provide important information about the impact of STR on NCF and PRO metrics. Moreover, they provide compelling data for a potentially novel endpoint for GBM clinical trials.
Disclosures
This project was supported by grants U10CA21661, U10CA180868, U10CA180822, U10 CA37422, and UG1CA189867 from the National Cancer Institute, Merck & Co, and Genentech. This publication was supported by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health (NIH), through Grant Number UL1TR001436. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
Dr Wefel reports consulting roles with Genentech, Roche, Bayer, VFW, Angiochem, Juno, Novocure, Cogstate, and Vanquish Oncology. Dr Wendland reports employment at US Oncology. Dr Chao reports honoraria from Zeiss and Abbvie, and travel expenses and speaking fees from Varian. Dr Roof reports Amgen, Merck, and Halyard Health Inc stock. Dr Mehta reports a leadership role and stock/ownership interest with Pharmacyclics, consulting roles with Varian medical Systems, Agenus, Insys Therapeutics, Remedy Pharmaceuticals, IBA, AstraZeneca, and Oncoceutics, research funding to his institution from Novocure and Novelos, a patent, and is a member of a data safety monitoring board for Monteris Medical. Dr Movsas reports research funding to institution from Varian Inc and Philips Inc, patents, and travel expenses from Varian Inc. Dr Brachman reports a patent or intellectual property interest in GammaTile. Dr Hall reports institutional research support from Elekta and grant funding from the American Cancer Society.
Supplementary Material
Notes
This study has been presented as an oral presentation at the 2014 Society of Neuro-oncology meeting in Miami, Florida, November 13 to 16, 2014, and the 2015 American Society of Radiation Oncology (ASTRO) meeting in San Antonio, Texas, October 18 to 21, 2015, both as oral presentations.
Supplemental Digital Content 1. Table. Change from Baseline to 6 wk for NCF for RTOG 0525 and RTOG 0825.
Supplemental Digital Content 2. Table. Change from Baseline to 6 wk for MDASI-BT for RTOG 0525 and RTOG 0825 Arm 1.
Supplemental Digital Content 3. Table. Change from Baseline to 6 wk for EORTC QLQ C30/BN20 for RTOG 0525 and RTOG 0825 Arm 1
Supplemental Digital Content 4. Table. Change from Baseline to 22 wk for NCF for RTOG 0525 and RTOG 0825 Arm 1.
Supplemental Digital Content 5. Table. Change from Baseline to 22 wk for MDASI-BT for RTOG 0525 and RTOG 0825 Arm 1.
Supplemental Digital Content 6. Table. Change from Baseline to 22 wk for EORTC QLQ C30/BN20 for RTOG 0525 and RTOG 0825 Arm 1.
Supplemental Digital Content 7. Table. Change from Baseline to 22 wk for NCF for RTOG 0825 Control vs Treatment.
Supplemental Digital Content 8. Table. Change from Baseline to 22 wk for MDASI-BT for RTOG 0825 Control vs Treatment.
Supplemental Digital Content 9. Table. Change from Baseline to 22 wk for EORTC QLQ C30/BN20 for RTOG 0825 Control vs Treatment.
REFERENCES
- 1. Dolecek TA, Propp JM, Stroup NE, Kruchko C. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2005-2009. Neuro Oncol. 2012;14(suppl 5):v1–v49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Stupp R, Mason WP, van den Bent MJ et al.. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352(10):987–996. [DOI] [PubMed] [Google Scholar]
- 3. Gilbert MR, Dignam JJ, Armstrong TS et al.. A randomized trial of bevacizumab for newly diagnosed glioblastoma. N Engl J Med. 2014; 370(8):699–708. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Gilbert MR, Wang M, Aldape KD et al.. Dose-dense temozolomide for newly diagnosed glioblastoma: a randomized phase III clinical trial. J Clin Oncol. 2013;31(32):4085–4091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Barone DG, Lawrie TA, Hart MG, Image guided surgery for the resection of brain tumours. Cochrane Database Syst Rev. 2014;1: Jan 28;(1):CD009685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Sanai N, Polley MY, McDermott MW, Parsa AT, Berger MS. An extent of resection threshold for newly diagnosed glioblastomas. J Neurosurg. 2011;115(1):3–8. [DOI] [PubMed] [Google Scholar]
- 7. Lacroix M, Abi-Said D, Fourney DR et al.. A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival. J Neurosurg. 2001;95(2):190–198. [DOI] [PubMed] [Google Scholar]
- 8. Pope WB, Sayre J, Perlina A, Villablanca JP, Mischel PS, Cloughesy TF. MR imaging correlates of survival in patients with high-grade gliomas. AJNR Am J Neuroradiol. 2005;26(10):2466–2474. [PMC free article] [PubMed] [Google Scholar]
- 9. Marko NF, Weil RJ, Schroeder JL, Lang FF, Suki D, Sawaya RE. Extent of resection of glioblastoma revisited: personalized survival modeling facilitates more accurate survival prediction and supports a maximum-safe-resection approach to surgery. J Clin Oncol. 2014;32(8):774–782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Jakola AS. Unsgard G, Solheim O. Quality of life in patients with intracranial gliomas: the impact of modern image-guided surgery. J Neurosurg. 2011;114(6):1622–1630. [DOI] [PubMed] [Google Scholar]
- 11. Sagberg LM, Solheim O, Jakola AS. Quality of survival the 1st year with glioblastoma: a longitudinal study of patient-reported quality of life. J Neurosurg. 2016;124(4):989–997. [DOI] [PubMed] [Google Scholar]
- 12. Budrukkar A, Jalali R, Dutta D et al.. Prospective assessment of quality of life in adult patients with primary brain tumors in routine neurooncology practice. J Neurooncol. 2009;95(3):413–419. [DOI] [PubMed] [Google Scholar]
- 13. Johnson DR, Sawyer AM, Meyers CA, O’Neill BP, Wefel JS. Early measures of cognitive function predict survival in patients with newly diagnosed glioblastoma. Neuro Oncol. 2012;14(6):808–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Armstrong TS, Wefel JS, Wang M et al.. Net clinical benefit analysis of radiation therapy oncology group 0525: a phase III trial comparing conventional adjuvant temozolomide with dose-intensive temozolomide in patients with newly diagnosed glioblastoma. J Clin Oncol. 2013;31(32):4076–4084. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Wefel JS, Pugh S, Armstrong TS, Gilbert MR, Won MM. Neurocognitive function (NCF) outcomes in patients with glioblastoma (GBM) enrolled in RTOG 0825. J Clin Oncol. 2013;31(15):2004–2004.23630218 [Google Scholar]
- 16. Ferrara N, Hillan KJ, Novotny W. Bevacizumab (Avastin), a humanized anti-VEGF monoclonal antibody for cancer therapy. Biochem Biophys Res Commun. 2005;333(2):328–335. [DOI] [PubMed] [Google Scholar]
- 17. Jain RK. Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science. 2005;307(5706):58–62. [DOI] [PubMed] [Google Scholar]
- 18. Efficace F, Bottomley A. Health related quality of life assessment methodology and reported outcomes in randomised controlled trials of primary brain cancer patients. Eur J Cancer. 2002;38(14):1824–1831. [DOI] [PubMed] [Google Scholar]
- 19. Taphoorn MJ, Claassens L, Aaronson NK et al.. An international validation study of the EORTC brain cancer module (EORTC QLQ-BN20) for assessing health-related quality of life and symptoms in brain cancer patients. Eur J Cancer. 2010;46(6):1033–1040. [DOI] [PubMed] [Google Scholar]
- 20. Armstrong TS, Mendoza T, Gning I et al.. Validation of the M.D. Anderson Symptom Inventory Brain Tumor Module (MDASI-BT). J Neurooncol. 2006;80(1):27–35. [DOI] [PubMed] [Google Scholar]
- 21. Shapiro AM, Benedict RH, Schretlen D, Brandt J. Construct and concurrent validity of the Hopkins Verbal Learning Test – Revised. Clin Neuropsychol. 1999;13(3):348–358. [DOI] [PubMed] [Google Scholar]
- 22. Tombaugh TN. Trail Making Test A and B: normative data stratified by age and education. Arch Clin Neuropsychol. 2004;19(2):203–214. [DOI] [PubMed] [Google Scholar]
- 23. Ruff RM, Light RH, Parker SB, Levin HS. Benton controlled Oral Word Association Test: reliability and updated norms. Arch Clin Neuropsychol. 1996;11(4):329–338. [PubMed] [Google Scholar]
- 24. Stupp R, Taillibert S, Kanner AA et al.. Maintenance therapy with tumor-treating fields plus temozolomide vs temozolomide alone for glioblastoma. JAMA. 2015;314(23):2535–2543. [DOI] [PubMed] [Google Scholar]
- 25. Meyers CA, Hess KR. Multifaceted end points in brain tumor clinical trials: cognitive deterioration precedes MRI progression. Neuro Oncol. 2003;5(2):89–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Pirzkall A, McGue C, Saraswathy S et al.. Tumor regrowth between surgery and initiation of adjuvant therapy in patients with newly diagnosed glioblastoma. Neuro Oncol. 2009;11(6):842–852. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.