An extent of resection threshold for recurrent glioblastoma and its risk for neurological morbidity

Clinical article

Full access

Object

Despite improvements in the medical and surgical management of patients with glioblastoma, tumor recurrence remains inevitable. For recurrent glioblastoma, however, the clinical value of a second resection remains uncertain. Specifically, what proportion of contrast-enhancing recurrent glioblastoma tissue must be removed to improve overall survival and what is the neurological cost of incremental resection beyond this threshold?

Methods

The authors identified 170 consecutive patients with recurrent supratentorial glioblastomas treated at the Barrow Neurological Institute from 2001 to 2011. All patients previously had a de novo glioblastoma and following their initial resection received standard temozolomide and fractionated radiotherapy.

Results

The mean clinical follow-up was 22.6 months and no patient was lost to follow-up. At the time of recurrence, the median preoperative tumor volume was 26.1 cm3. Following re-resection, median postoperative tumor volume was 3.1 cm3, equating to an 87.4% extent of resection (EOR). The median overall survival was 19.0 months, with a median progression-free survival following re-resection of 5.2 months. Using Cox proportional hazards analysis, the variables of age, Karnofsky Performance Scale (KPS) score, and EOR were predictive of survival following repeat resection (p = 0.0001). Interestingly, a significant survival advantage was noted with as little as 80% EOR. Recursive partitioning analysis validated these findings and provided additional risk stratification at the highest levels of EOR. Overall, at 7 days after surgery, a deterioration in the NIH stroke scale score by 1 point or more was observed in 39.1% of patients with EOR ≥ 80% as compared with 16.7% for those with EOR < 80% (p = 0.0049). This disparity in neurological morbidity, however, did not endure beyond 30 days postoperatively (p = 0.1279).

Conclusions

For recurrent glioblastomas, an improvement in overall survival can be attained beyond an 80% EOR. This survival benefit must be balanced against the risk of neurological morbidity, which does increase with more aggressive cytoreduction, but only in the early postoperative period. Interestingly, this putative EOR threshold closely approximates that reported for newly diagnosed glioblastomas, suggesting that for a subset of patients, the survival benefit of microsurgical resection does not diminish despite biological progression.

Abbreviations used in this paper:EOR = extent of resection; ICC = intraclass correlation coefficient; KPS = Karnofsky Performance Scale; NIHSS = NIH Stroke Scale; RPA = recursive partitioning analysis.

Abstract

Object

Despite improvements in the medical and surgical management of patients with glioblastoma, tumor recurrence remains inevitable. For recurrent glioblastoma, however, the clinical value of a second resection remains uncertain. Specifically, what proportion of contrast-enhancing recurrent glioblastoma tissue must be removed to improve overall survival and what is the neurological cost of incremental resection beyond this threshold?

Methods

The authors identified 170 consecutive patients with recurrent supratentorial glioblastomas treated at the Barrow Neurological Institute from 2001 to 2011. All patients previously had a de novo glioblastoma and following their initial resection received standard temozolomide and fractionated radiotherapy.

Results

The mean clinical follow-up was 22.6 months and no patient was lost to follow-up. At the time of recurrence, the median preoperative tumor volume was 26.1 cm3. Following re-resection, median postoperative tumor volume was 3.1 cm3, equating to an 87.4% extent of resection (EOR). The median overall survival was 19.0 months, with a median progression-free survival following re-resection of 5.2 months. Using Cox proportional hazards analysis, the variables of age, Karnofsky Performance Scale (KPS) score, and EOR were predictive of survival following repeat resection (p = 0.0001). Interestingly, a significant survival advantage was noted with as little as 80% EOR. Recursive partitioning analysis validated these findings and provided additional risk stratification at the highest levels of EOR. Overall, at 7 days after surgery, a deterioration in the NIH stroke scale score by 1 point or more was observed in 39.1% of patients with EOR ≥ 80% as compared with 16.7% for those with EOR < 80% (p = 0.0049). This disparity in neurological morbidity, however, did not endure beyond 30 days postoperatively (p = 0.1279).

Conclusions

For recurrent glioblastomas, an improvement in overall survival can be attained beyond an 80% EOR. This survival benefit must be balanced against the risk of neurological morbidity, which does increase with more aggressive cytoreduction, but only in the early postoperative period. Interestingly, this putative EOR threshold closely approximates that reported for newly diagnosed glioblastomas, suggesting that for a subset of patients, the survival benefit of microsurgical resection does not diminish despite biological progression.

Glioblastoma is the most common primary malignant brain tumor in adults, and the prognosis for patients is dismal. The standard treatment for newly diagnosed glioblastomas is microsurgical resection followed by concomitant chemotherapy and radiation therapy.18,25,28,31 Unfortunately, despite years of refinement, this multimodal approach still leads to a median survival time of 12–15 months for most patients.13,22,31 Nevertheless, microsurgical resection continues to be the first treatment stage for nearly all patients with newly diagnosed glioblastoma, particularly in light of recent evidence demonstrating a survival benefit associated with greater extent of resection (EOR).11 In the largest study to date, volumetric analysis of 500 patients with newly diagnosed glioblastoma suggests that an EOR as low as 78% is associated with improved overall survival.25 Incremental improvement in overall survival is observed beyond this margin as well, even at the highest intervals of resection.18,25 Taken together, these data emphasize the utility of both subtotal and gross-total resections for patients with newly diagnosed glioblastoma in the modern era.

However, despite improvements in the medical and surgical management of patients with glioma, glioblastoma recurrence remains inevitable. For many patients, quality of life at the time of tumor recurrence is high and thus repeat resection is an increasingly common consideration.5 Unlike newly diagnosed glioblastomas, however, our understanding of the morbidity profile and survival benefit of re-resection remains incomplete. While several recent studies have correlated a survival benefit with recurrent glioblastoma gross-total resection, most are limited by small sample sizes, categorical measurements of the EOR, and heterogeneity in adjuvant treatment regimens.1,3,5,7,10,12,14,20,23 Furthermore, survival analyses examining the EOR without consideration of neurological morbidity likely provide an incomplete assessment of the intervention and its true clinical impact. In addition, no study to date has investigated whether an EOR threshold for recurrent glioblastoma correlates with patient survival, or how this threshold compares to the threshold for newly diagnosed tumors. We have therefore attempted to quantify the effect of the EOR on recurrent glioblastoma patient survival in the context of neurological morbidity and have attempted to define a possible threshold for tumor cytoreduction.

Methods

Patient Population

We identified 170 consecutive adult patients who underwent planned resection of their recurrent supratentorial glioblastoma at Barrow Neurological Institute between 2001 and 2011. All patients demonstrated MRI evidence of new tumor growth at the time of repeat surgery. Central neuropathology review confirmed a diagnosis of WHO Grade IV glioma at both initial and repeat surgeries, based on WHO guidelines.19 All patients with secondary glioblastoma were excluded from the analysis, as were those with evidence of gliosarcoma. Patients who did not undergo standard (Stupp regimen) adjuvant chemotherapy and radiation therapy28 following resection of their initial tumor were excluded, as were patients enrolled in clinical trials that modified this protocol. However, 52 (30.6%) of 170 patients received bevacizumab, and 72 (42.4%) received carmustine wafers during their treatment course.

Data Collection

Clinical, radiographic, and outcome data were collected from inpatient and outpatient records, telephone interviews, and the national Social Security Death Index. Characteristics identified for each patient included age and Karnofsky Performance Scale (KPS) score at the time of repeat surgery, pre- and postoperative tumor volumes, volumetric EOR, adjuvant therapy, tumor location, eloquence of tumor location, and time intervals from initial surgery to repeat surgery and from repeat surgery to further tumor progression or death. Eloquence was defined by radiological tumor location;2,16,24 in patients undergoing motor and/or speech mapping during resection, eloquence was confirmed by the presence of a positive intraoperative map.

A neurological examination was performed in all patients by an attending neurosurgeon at preoperative and postoperative visits. A motor deficit was defined as decreased strength, while a language deficit was defined as any combination of receptive and/or expressive aphasia. For assessment of acute changes in neurological function, the NIH Stroke Scale (NIHSS) score4,9,32 was adapted as an outcome parameter. The NIHSS score assesses 15 neurological functions, grading the severity of impairment for each function individually, ranging from 0 (best) to 42 (worst) points. The score was assigned preoperatively and at approximately 7 days after and 30 days after surgery. The institutional review board of St. Joseph's Hospital and Medical Center approved this retrospective study.

Tumor Volume and EOR Calculation

The EOR for each case was determined by comparing MR images obtained before surgery to those obtained within 48 hours after surgery. A blinded, retrospective 3D volumetric analysis of pre- and postoperative MR images was performed by a neuroradiologist. Manual segmentation was performed with region-of-interest measurements of tumor volumes (in cm3) based on contrast-enhancing tissue observed on T1-weighted MRI. Extent of resection was calculated as follows: (preoperative tumor volume − postoperative tumor volume)/preoperative tumor volume. Volume measurements were obtained without consideration of patient outcome.

To assess the reliability of tumor volumes, pre- and postoperative tumor volumetric measurements were repeated in a population of 20 randomly selected patients from the original study cohort. The repeat measurements were made by a neurosurgeon (N.S.) blinded to previous volumetric measurements and to patient outcome, using the same methodology described above. Cronbach's α29 and the intraclass correlation coefficient (ICC)30 were used to assess the reliability of measurements between observers. For interpretation, a Cronbach's α ≥ 0.9 may be considered excellent internal consistency,17 whereas an ICC ≥ 0.75 indicates excellent reliability.8

Statistical Analysis

Age, EOR, KPS scores, NIHSS scores, and tumor volumes were analyzed as continuous variables. Medians and ranges were calculated for continuous variables to summarize these characteristics. Counts and percentages were defined for categorical variables. The Wilcoxon rank-sum test for continuous or ordinal variables was used to compare patient and treatment characteristics among groups. The Kendall τ correlation was used to assess the correlation between 2 continuous variables.

To evaluate the prognostic value of the variables under consideration, a Cox proportional hazards model was used to identify all EOR categories associated with improved survival. Variables that were significant at the α = 0.2 level in the univariate analysis were then entered into a multivariate model for consideration. The forward stepwise selection technique was then used to select the final variables to retain. Only variables that were statistically significant at the p = 0.01 level were included in the final model. Kaplan-Meier curves were then constructed to summarize the relative impact of each EOR category, and to identify an EOR threshold.

A second approach with recursive partitioning analysis (RPA) was then used to identify the combined prognostic category associated with the maximal impact on overall glioblastoma survival. We followed the method of exponential scaling, in which the survival time was prescaled to fit a parametric exponential model. Ten-fold cross-validation was used. The maximum-size tree for which the complexity parameter exceeds the p < 0.01 threshold was chosen as the final tree. Once the tree was selected, the log-rank test with a significance level of p < 0.01 was used to confirm the difference for each split identified by the tree. Any split that did not satisfy this criterion was discarded. The terminal nodes were then compared again, using the log-rank test with the p < 0.01 criterion. Final nodes that did not meet the criterion were combined.

Results

Clinical and Surgical Demographics

One hundred and seventy patients who met the abovementioned inclusion and exclusion criteria were identified (Table 1). The mean patient age was 55.2 years (range 19–78 years); 105 patients (61.8%) were male; and patients presented with a median KPS score of 80 (range 30–100). The mean clinical follow-up duration was 22.6 months (range 3.7–103.7 months), with no patient unaccounted. The most common area of tumor infiltration was the frontal lobe (n = 90, 52.9%), and nearly half of the tumors (n = 84, 49.4%) occupied eloquent territory. All patients received standard temozolomide and fractionated radiotherapy after initial resection.26 A median interval of 8.9 months (range 1.1–93.1 months) elapsed between initial and repeat resections. The median preoperative tumor volume upon recurrence was 26.1 cm3 (range 1.0–160.2 cm3), and all patients underwent image-guided microsurgical resection of their tumor recurrence. The median postoperative tumor volume was 3.1 cm3 (range 0–73.6 cm3), equating to an 87.4% median EOR (range 11.1%–100% EOR). Tumors located in the frontal and temporal lobes were more likely to be completely resected. The median overall survival was 19.0 months (range 3.7–103.7 months), which represented a more favorable profile as compared with an unselected newly diagnosed glioblastoma population (Fig. 1).25 The median progression-free survival after re-resection was 5.2 months.

TABLE 1:

Preoperative characteristics of patients who underwent microsurgical resection of recurrent glioblastomas

CharacteristicTotal≥80% EOR<80% EORp Value
patients17011060
mean age ± SD (yrs)55.2 ± 11.9754.9 ± 12.5355.6 ± 10.950.9287
sex (%)0.0964
 male105 (61.8)72 (65.5)33 (55)
 female65 (38.2)38 (34.5)27 (45)
KPS score (%)0.6101
 10012 (7.1)9 (8.2)3 (5)
 9028 (16.5)21 (19.1)7 (11.7)
 8058 (34.1)39 (35.5)19 (31.7)
 7038 (22.4)23 (20.9)15 (25.0)
 ≤6034 (20.0)18 (16.4)16 (26.7)
 median (range)80 (30–100)80 (30–100)70 (30–100)
NIHSS score distribution (%)0.0540
 052 (30.6)37 (33.6)15 (25.0)
 121 (12.4)15 (13.6)6 (10.0)
 241 (24.1)27 (24.5)14 (23.3)
 318 (10.6)11 (10.0)7 (11.7)
 414 (8.2)9 (8.2)5 (8.3)
 510 (5.9)4 (3.6)6 (10.0)
 65 (2.9)2 (1.8)3 (5.0)
 73 (1.8)2 (1.8)1 (1.7)
 83 (1.8)2 (1.8)1 (1.7)
 92 (1.2)1 (0.9)1 (1.7)
 101 (0.6)01 (1.7)
 median (range)2 (0–10)2 (0–9)2 (0–10)
preop tumor volume (cm3)0.1366
 mean ± SD34.5 ± 28.0532.5 ± 27.2838.12 ± 29.28
 median (range)26.1 (1.0–160.2)23.8 (1.0–150.8)28.1 (1.6–160.2)
tumor location (%)
 frontal90 (52.9)51 (46.4)39 (65.0)0.0223
 temporal76 (44.7)42 (38.2)34 (56.7)0.0209
 parietal57 (33.5)34 (30.9)23 (38.3)0.2950
 occipital38 (22.4)27 (24.5)11 (18.3)0.4700
eloquent tumor location (%)0.0520
 yes84 (49.4)48 (43.6)36 (60.0)
 no86 (50.6)62 (56.4)24 (40.0)
Fig. 1.
Fig. 1.

Graph of overall survival of recurrent versus newly diagnosed glioblastoma. Kaplan-Meier curves show overall survival of the current study's recurrent glioblastoma patient population, and an unselected newly diagnosed glioblastoma patient population from a previous study.25

Reliability of Tumor Volumetric Measurements

Preoperative and postoperative tumor volumetric measurements were repeated in 20 randomly selected patients from the original study cohort by a blinded neurosurgeon (N.S.). Based on these measurements, Cronbach's α for pre- and postoperative tumor volumes was 0.910 and 0.928, respectively. The ICC for pre- and postoperative tumor volumes was 0.832 (95% CI 0.726–0.904) and 0.866 (95% CI 0.692–0.945), respectively. These results suggest excellent interobserver reliability of tumor volumetric measurements.

Aggressive Resection and Neurological Morbidity

At the time of recurrence, 56 patients (33%) demonstrated a focal motor deficit and 52 patients (31%) demonstrated a focal language deficit on preoperative examination. Following microsurgical resection, these new or worsened deficits developed at 1 week in 19% and 13%, respectively, but improved to 15% and 9% at 4 weeks. The mean deterioration (± SD) in the NIHSS score at 7 days after surgery was 1.17 ± 2.61 in those with EOR ≥ 80% compared with 0.53 ± 2.07 in those with EOR < 80%. However, at the 95% EOR interval, the mean deterioration in the NIHSS score at 7 days after surgery was 1.27 ± 2.71 (EOR ≥ 95%) compared with 0.82 ± 2.33 (EOR < 95%). These statistically significant differences were more disparate among patients with eloquent tumors: at every EOR interval, a higher proportion of patients demonstrated postoperative deterioration (based on NIHSS score) if their tumors were in an eloquent location (Table 2). One month after repeat resection, the mean deterioration in NIHSS scores at the 80% EOR interval was 0.71 ± 2.27 for those with EOR ≥ 80% compared with 0.30 ± 2.02 for those with EOR < 80%; the mean deterioration in NIHSS scores at the 95% EOR interval was 0.77 ± 2.32 for those with EOR ≥ 95% compared with 0.51 ± 2.14 for those with EOR < 95%. Interestingly, by 30 days postoperatively, EOR was no longer predictive of new or worsened neurological deficits (Table 3). Similarly, at 7 days after surgery, a deterioration in the NIHSS score by 1 point or more was observed in 39.1% of patients with EOR ≥ 80% compared with 16.7% for those with EOR < 80% (p = 0.0049). At 30 days, however, the distinction (25.5% vs 15.0%, respectively) was also nonsignificant (p = 0.1279, Table 3).

TABLE 2:

Deterioration of the NIHSS score by at least 1 point at 7 days compared with preoperative score

Eloquent Tumor LocationDeterioration %Deterioration %
≥80% EOR<80% EORp Value≥95% EOR<95% EORp Value
yes50.025.00.039168.430.80.1072
no29.08.30.026931.019.30.0964
overall39.116.70.004945.825.40.0207
TABLE 3:

Deterioration of the NIHSS score by at least 1 point compared with preoperative score

TimeDeterioration %Deterioration %
≥80% EOR<80% EORp Value≥95% EOR<95% EORp Value
7 days postop39.116.70.004945.825.40.0207
30 days postop25.515.00.127931.318.00.2551

Extent of Resection Threshold for Recurrent Glioblastoma

Univariate Cox proportional hazards model analysis was used to examine each collected variable and identify those that were statistically significant at the p = 0.20 level. Those variables were age (p < 0.0001), KPS score (p < 0.0001), EOR (p = 0.001), and postoperative tumor volume (p = 0.001). Based on this initial survey, a multivariate Cox proportional hazards analysis was built using a forward stepwise selection technique, with the final model retaining only variables significant at the p = 0.01 level. This final analysis designated age (p = 0.0001), KPS score (p = 0.001), and EOR (p = 0.005) as predictors of overall survival among patients with recurrent glioblastoma (Table 4).

TABLE 4:

Multivariate Cox proportional hazards analysis of overall survival

VariableHazard Ratio (95% CI)p Value
age1.04 (1.03–1.05)0.0001
KPS score0.96 (0.95–0.98)0.001
EOR0.97 (0.96–0.98)0.005

To explore the relative impact of subtotal resections, serial Kaplan-Meier survival curves were generated at 1% EOR intervals. A significant survival advantage was noted with as little as 80% EOR, which was associated with a 19.2-month median survival. An EOR ≥ 81% equated to a 20-month median survival, whereas an EOR of ≥ 97% equated to a 30.0-month median survival. Stepwise improvement in overall survival was evident even within the 95%–100% range (p < 0.001), suggesting a role for additional risk-group stratification within the highest EOR subgroups (Fig. 2).

Fig. 2.
Fig. 2.

Line graph of overall survival stratified by EOR. Kaplan-Meier curves show overall survival in a patient population that underwent resection of recurrent glioblastoma at EOR levels of ≤ 100%, ≤ 90%, ≤ 80%, or ≤ 70%.

Risk Stratification of Recurrent Glioblastoma Surgery and Impact on Overall Survival

An RPA was constructed to validate our above statistical analysis, identify the EOR category with the largest measurable effect on overall survival, and risk-stratify patients within this EOR category. The results matched the multivariate Cox proportional hazards analysis, identifying patient age, KPS score, EOR, and tumor volume as each predictive of overall survival (p < 0.0001). Also, as indicated by the Kaplan-Meier survival curve analysis, an EOR ≥ 97% had the largest impact on overall survival, with the 30 patients in this EOR category demonstrating a median overall survival of 30.0 months. Within this cohort, RPA branching defined 4 risk groups with successively worse outcomes: Group 1 (EOR ≥ 97%, age ≤ 41), Group 2 (EOR ≥ 97%, age 41–66), Group 3 (EOR ≥ 97%, age ≥ 67, preoperative volume < 12.1 cm3), and Group 4 (EOR ≥ 97%, age ≥ 67, preoperative volume ≥ 12.1 cm3; Fig. 3). Thus, for patients with EOR of 97%–100%, these statistically significant stratifications correspond to low-, low-moderate-, moderate-high-, and high-risk subgroups. The respective 6-month overall survival rates in these 4 groups were 100%, 77%, 61%, and 53%.

Fig. 3.
Fig. 3.

Recursive partitioning analysis results identified each of the following as predictive of overall survival (OS) with recurrent glioblastoma: patient age, KPS score, tumor volume, and EOR (p < 0.0001). These 4 distinct risk groups were each predictive of outcome. In the interest of brevity, additional stratification based on KPS score was omitted from this figure.

Discussion

Beyond establishing a diagnosis and decompressing the mass effect, the clinical value of microsurgical resection for patients with recurrent glioblastoma remains controversial. Our findings suggest that both subtotal and gross-total resection can benefit re-resection candidates, even at an EOR as low as 80%. This cytoreduction does, however, come at some neurological cost to the patient. Significant differences in neurological morbidity, based on NIHSS scores, were observed with more extensive resection, particularly within the early postoperative period. At the highest levels of EOR, these deficits had a greater tendency to remain permanent. Furthermore, patients at particular risk for deterioration were those with preexisting neurological symptoms and tumors in eloquent locations. Collectively, these observations may help identify patients with recurrent glioblastoma at the greatest risk for permanent neurological deficit and thus allow for tailoring of their surgical strategies.

Several prior studies compared patients with recurrent glioblastoma who did or did not undergo reoperation.1,10,14,20,23 While these studies suggested a survival benefit with reoperation, their small sample sizes (mean 40 patients, range 18–65 patients) and heterogeneity in adjuvant therapy after initial diagnosis (range 34%–100%) hampered firm conclusions. More recently, efforts to evaluate gross-total versus subtotal resection for recurrent glioblastomas have also suggested an overall survival benefit associated with gross-total resection.3,5 The generalizability of these studies to the modern era, however, is also limited, as no study included concomitant assessment of neurological morbidity alongside EOR, nor excluded patients who underwent nonconventional or experimental adjuvant therapy regimens at the time of initial diagnosis. The critical importance of adjuvant therapy homogeneity when evaluating EOR for recurrent glioblastoma is evidenced by a recent large-scale meta-analysis of North American Brain Tumor Consortium Phase II trials that demonstrated no difference in overall survival when comparing surgical and nonsurgical patients with glioblastoma recurrence,6 an observation likely influenced by sample heterogeneity.

Although distinguishing gross-total versus subtotal resection provides a basic framework for neurosurgical decision-making, the routine assignment of 95% as the EOR threshold separating these categories is arbitrary and oversimplifies the intraoperative challenges of the neurosurgical oncologist. Specifically, the proximity of most recurrent glioblastomas to functional pathways,3,27 as well as the critical importance of incurred neurological morbidity to overall survival,21 emphasizes the value of determining a minimum EOR threshold. For newly diagnosed glioblastomas, an initial effort to address this question emerged from a 500-patient study of newly diagnosed glioblastomas uniformly treated with adjuvant chemotherapy and radiation therapy, suggesting that 78% EOR is the minimum at which a survival benefit is first evident.25 In that study, the median overall survival was 12.2 months, with 78% EOR equating to a 12.5-month overall survival and EOR ≥ 97% equating to a 15.8-month overall survival. Similarly, in the current study, we explored the value of a subtotal resection among patients with recurrent glioblastoma using a post hoc analysis strategy based on the Kaplan-Meier method and found that an EOR as low as 80% may be associated with a survival benefit. Because this analysis was exploratory, it did not adjust for other putative prognostic factors due to sample size constraints and therefore cannot rule out that the survival differences observed at this threshold may be driven by variables other than EOR.

Our experience with patients with recurrent glioblastoma compared with a recent EOR analysis of newly diagnosed glioblastomas25 demonstrates a higher median overall survival for all patients (19.0 vs 12.2 months, respectively), as well as for those with very aggressive reresection (that is, for EOR ≥ 97%: 30.0 vs 15.8 months, respectively). This comparatively favorable survival profile may reflect the biological heterogeneity of newly diagnosed glioblastomas, with patients eligible for re-resection likely harboring comparatively indolent versions of the tumor. Interestingly, this self-selection did not substantially alter the putative minimum EOR threshold for survival benefit (78% for newly diagnosed glioblastoma25 vs 80% for recurrent glioblastoma in the current study) and raises the possibility that, despite the molecular and genetic changes that routinely evolve with glioblastoma recurrence,31 new biological features of glioblastoma do not affect their responsiveness to microsurgical intervention. Future studies could help delineate whether distinct genetic and/or epigenetic features of glioblastomas determine their likelihood to respond favorably to cytoreduction.

This retrospective study was undertaken to define the clinical value of EOR, both in terms of neurological morbidity and overall survival, for patients with glioblastoma. The statistical limitations of such a retrospective analysis are well known15,18,25 and cannot be completely controlled for with any statistical model. Nevertheless, we attempted to homogenize our data set by studying a large sample size, while excluding patients with atypical adjuvant treatment regimens, secondary glioblastomas, and multiple prior resections. We agree with prior reports that, without randomization, equal distribution of known and unknown confounders cannot be assured. However, for patients with recurrent glioblastoma, we believe such a study is not beyond reach. Accordingly, we have recently initiated a Phase III randomized trial at Barrow Neurological Institute comparing re-resection to biopsy for patients with recurrent glioblastoma.

Conclusions

The clinical value of the EOR for recurrent glioblastoma remains a topic of debate, particularly for incomplete resections. Our findings, based upon a large and relatively homogeneous recurrent de novo glioblastoma patient population, indicate that an EOR as low as 80% may impact survival and that this trend continues even at the highest levels of resection. We also quantify, however, the added risk of re-resection, as greater cytoreduction is accompanied by a rising incidence of transient neurological morbidity. Attaining an EOR beyond this threshold, as well as preserving functional pathways to maximize quality of life, should be of critical concern to the neurosurgeon treating patients with recurrent glioblastoma.

Acknowledgment

We would like to acknowledge John P. Karis, M.D., for performing the tumor volumetric analysis.

Disclosure

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author contributions to the study and manuscript preparation include the following. Conception and design: Sanai. Acquisition of data: Oppenlander, Wolf, Snyder, Bina. Analysis and interpretation of data: Sanai, Oppenlander, Snyder, Ashby, Brachman. Drafting the article: Sanai, Oppenlander. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Sanai. Statistical analysis: Sanai, Wilson. Administrative/technical/material support: Sanai, Oppenlander, Coons, Nakaji, Porter, Smith, Spetzler. Study supervision: Sanai, Oppenlander.

References

If the inline PDF is not rendering correctly, you can download the PDF file here.

Article Information

Address correspondence to: Nader Sanai, M.D., c/o Neuroscience Publications, Barrow Neurological Institute, St. Joseph's Hospital and Medical Center, 350 W. Thomas Rd., Phoenix, AZ 85013. email: neuropub@dignityhealth.org.

Please include this information when citing this paper: published online January 31, 2014; DOI: 10.3171/2013.12.JNS13184.

© AANS, except where prohibited by US copyright law.

Headings

Figures

  • View in gallery

    Graph of overall survival of recurrent versus newly diagnosed glioblastoma. Kaplan-Meier curves show overall survival of the current study's recurrent glioblastoma patient population, and an unselected newly diagnosed glioblastoma patient population from a previous study.25

  • View in gallery

    Line graph of overall survival stratified by EOR. Kaplan-Meier curves show overall survival in a patient population that underwent resection of recurrent glioblastoma at EOR levels of ≤ 100%, ≤ 90%, ≤ 80%, or ≤ 70%.

  • View in gallery

    Recursive partitioning analysis results identified each of the following as predictive of overall survival (OS) with recurrent glioblastoma: patient age, KPS score, tumor volume, and EOR (p < 0.0001). These 4 distinct risk groups were each predictive of outcome. In the interest of brevity, additional stratification based on KPS score was omitted from this figure.

References

1

Barker FG IIChang SMGutin PHMalec MKMcDermott MWPrados MD: Survival and functional status after resection of recurrent glioblastoma multiforme. Neurosurgery 42:7097231998

2

Berger MSCohen WAOjemann GA: Correlation of motor cortex brain mapping data with magnetic resonance imaging. J Neurosurg 72:3833871990

3

Bloch OHan SJCha SSun MZAghi MKMcDermott MW: Impact of extent of resection for recurrent glioblastoma on overall survival. Clinical article. J Neurosurg 117:103210382012

4

Brott TAdams HP JrOlinger CPMarler JRBarsan WGBiller J: Measurements of acute cerebral infarction: a clinical examination scale. Stroke 20:8648701989

5

Chaichana KLZadnik PWeingart JDOlivi AGallia GLBlakeley J: Multiple resections for patients with glioblastoma: prolonging survival. Clinical article. J Neurosurg 118:8128202013

6

Clarke JLEnnis MMYung WKChang SMWen PYCloughesy TF: Is surgery at progression a prognostic marker for improved 6-month progression-free survival or overall survival for patients with recurrent glioblastoma?. Neuro Oncol 13:111811242011

7

Dirks PBernstein MMuller PJTucker WS: The value of reoperation for recurrent glioblastoma. Can J Surg 36:2712751993

8

Fleiss JL: The Design and Analysis of Clinical Experiments New YorkWiley1986

9

Goldstein LBBertels CDavis JN: Interrater reliability of the NIH stroke scale. Arch Neurol 46:6606621989

10

Guyotat JSignorelli FFrappaz DMadarassy GRicci ACBret P: Is reoperation for recurrence of glioblastoma justified?. Oncol Rep 7:8999042000

11

Hardesty DASanai N: The value of glioma extent of resection in the modern neurosurgical era. Front Neurol 3:1402012

12

Harsh GR IVLevin VAGutin PHSeager MSilver PWilson CB: Reoperation for recurrent glioblastoma and anaplastic astrocytoma. Neurosurgery 21:6156211987

13

Hegi MEDiserens ACGorlia THamou MFde Tribolet NWeller M: MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352:99710032005

14

Helseth RHelseth EJohannesen TBLangberg CWLote KRønning P: Overall survival, prognostic factors, and repeated surgery in a consecutive series of 516 patients with glioblastoma multiforme. Acta Neurol Scand 122:1591672010

15

Hess KR: Extent of resection as a prognostic variable in the treatment of gliomas. J Neurooncol 42:2272311999

16

Kido DKLeMay MLevinson AWBenson WE: Computed tomographic localization of the precentral gyrus. Radiology 135:3733771980

17

Kline P: The Handbook of Psychological Testing ed 2LondonRoutledge2000

18

Lacroix MAbi-Said DFourney DRGokaslan ZLShi WDeMonte F: A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival. J Neurosurg 95:1901982001

19

Louis DNOhgaki HWiestler ODCavenee WKBurger PCJouvet A: The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114:971092007

20

Mandl ESDirven CMBuis DRPostma TJVandertop WP: Repeated surgery for glioblastoma multiforme: only in combination with other salvage therapy. Surg Neurol 69:5065092008

21

McGirt MJMukherjee DChaichana KLThan KDWeingart JDQuinones-Hinojosa A: Association of surgically acquired motor and language deficits on overall survival after resection of glioblastoma multiforme. Neurosurgery 65:4634702009

22

Mineo JFBordron ABaroncini MRamirez CMaurage CABlond S: Prognosis factors of survival time in patients with glioblastoma multiforme: a multivariate analysis of 340 patients. Acta Neurochir (Wien) 149:2452532007

23

Pinsker MLumenta C: Experiences with reoperation on recurrent glioblastoma multiforme. Zentralbl Neurochir 62:43472001

24

Pouratian NBookheimer SY: The reliability of neuroanatomy as a predictor of eloquence: a review. Neurosurg Focus 28:2E32010

25

Sanai NPolley MYMcDermott MWParsa ATBerger MS: An extent of resection threshold for newly diagnosed glioblastomas. Clinical article. J Neurosurg 115:382011

26

Sanai NSnyder LAHonea NJCoons SWEschbacher JMSmith KA: Intraoperative confocal microscopy in the visualization of 5-aminolevulinic acid fluorescence in low-grade gliomas. Clinical article. J Neurosurg 115:7407482011

27

Skirboll SSOjemann GABerger MSLettich EWinn HR: Functional cortex and subcortical white matter located within gliomas. Neurosurgery 38:6786851996

28

Stupp RMason WPvan den Bent MJWeller MFisher BTaphoorn MJ: Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 352:9879962005

29

Tavakol MDennick R: Making sense of Cronbach's alpha. Int J Med Ed 2:53552011

30

Weir JP: Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res 19:2312402005

31

Wen PYKesari S: Malignant gliomas in adults. N Engl J Med 359:4925072008

32

Williams LSYilmaz EYLopez-Yunez AM: Retrospective assessment of initial stroke severity with the NIH Stroke Scale. Stroke 31:8588622000

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 533 533 363
PDF Downloads 174 174 31
EPUB Downloads 0 0 0

PubMed

Google Scholar