Extent of Resection Thresholds in Molecular Subgroups of Newly Diagnosed Isocitrate Dehydrogenase–Wildtype Glioblastoma

Abstract

BACKGROUND AND OBJECTIVES:

Maximizing the extent of resection (EOR) improves outcomes in glioblastoma (GBM). However, previous GBM studies have not addressed the EOR impact in molecular subgroups beyond IDH1/IDH2 status. In the current article, we evaluate whether EOR confers a benefit in all GBM subtypes or only in particular molecular subgroups.

METHODS:

A retrospective cohort of newly diagnosed GBM isocitrate dehydrogenase (IDH)–wildtype undergoing resection were prospectively included in a database (n = 138). EOR and residual tumor volume (RTV) were quantified with semiautomated software. Formalin-fixed paraffin-embedded tumor tissues were analyzed by targeted next-generation sequencing. The association between recurrent genomic alterations and EOR/RTV was evaluated using a recursive partitioning analysis to identify thresholds of EOR or RTV that may predict survival. The Kaplan–Meier methods and multivariable Cox proportional hazards regression methods were applied for survival analysis.

RESULTS:

Patients with EOR ≥88% experienced 44% prolonged overall survival (OS) in multivariable analysis (hazard ratio: 0.56, P = .030). Patients with alterations in the TP53 pathway and EOR <89% showed reduced OS compared to TP53 pathway altered patients with EOR>89% (10.5 vs 18.8 months; HR: 2.78, P = .013); however, EOR/RTV was not associated with OS in patients without alterations in the TP53 pathway. Meanwhile, in all patients with EOR <88%, PTEN-altered had significantly worse OS than PTEN-wildtype (9.5 vs 15.4 months; HR: 4.53, P < .001).

CONCLUSION:

Our results suggest that a subset of molecularly defined GBM IDH-wildtype may benefit more from aggressive resections. Re-resections to optimize EOR might be beneficial in a subset of molecularly defined GBMs. Molecular alterations should be taken into consideration for surgical treatment decisions in GBM IDH-wildtype.

ABBREVIATIONS:

EOR

extent of resection

HR

hazard ratio

IDH

isocitrate dehydrogenase

KPS

Karnofsky performance status score

NGS

next-generation sequencing

OS

overall survival

RPA

recursive partitioning analysis

RTV

residual tumor volume.

Glioblastoma (GBM) is the most common malignant brain tumor in adults. The average annual age-adjusted incidence rates were 3.23 per 100 000 population in the United States from 2014 to 2018.¹ The current standard care for GBM is maximal safe resection followed by radiotherapy and concomitant chemotherapy with temozolomide.² Nevertheless, the treatment for GBM remains challenging, and the median patient overall survival (OS) in clinical trials is about 15 months.²

The poor prognosis of GBM is attributed to its highly invasive characteristic that makes curative surgical resection impossible. However, the importance of the extent of resection (EOR) for GBM has been discussed and proved for decades.^3-6 Growing evidence supports not only gross total resection of the enhanced lesion but also varying degrees of supratotal resection can contribute to increasing survival. Landmark studies demonstrated that certain thresholds of EOR were associated with a survival benefit.^7-9 In recent reports, minimizing residual tumor volume (RTV) has also been established as a prognostic factor and several RTV thresholds to improve survival have been established.^9-12

Survival outcomes widely vary in patients with GBM even after the same EOR. Genetic alterations, such as IDH1/IDH2 mutations and MGMT promoter methylation status, significantly affect patient survival.^13,14 Indeed, genetic makeup has become crucial for glioma classification, and IDH-mutant GBM name has been changed to Astrocytoma IDH-mutant Grade 4 to denote its different biology and prognosis.¹⁵ Recent articles stratified patients with GBM by IDH status and/or MGMT promoter methylation status and showed the different prognostic values of EOR within those molecular subgroups.^16-19 However, genetic or epigenetic features, other than MGMT promoter methylation status, that affect the efficacy of cytoreductive surgery for IDH-wildtype GBM are still unclear. We hypothesized that the outcome and role of surgical resection would be different among the various genomic alterations of GBM. Therefore, this study aimed to identify gene alterations in which maximal resection confers a benefit and to evaluate various cutoff thresholds of EOR and RTV associated with prolonged survival in molecular subgroups of IDH-wildtype GBM.

METHODS

Patients and Tumor Samples

Patients diagnosed with GBM IDH-wildtype and next-generation sequencing (NGS) data from 2009 to 2018 were selected from our glioma database. The following criteria of exclusion were used: the presence of IDH mutation, multifocal/multicentric, brainstem and/or cerebellum location, no resection (biopsy or imaging diagnosis), patients without preoperative and postoperative MRI, and no NGS data. Data for this study were collected from the electronic medical record and compiled using REDCap electronic data capture tools.^20,21 Data collected included age, sex, Karnofsky performance status score (KPS), diagnosis, tumor location, volumetric EOR, initial treatment strategy, recurrence, and date of death/last follow-up. Tumors were classified by a board-certified neuropathologist following the 2021 World Health Organization Classification of Tumors of the Central Nervous System.¹⁵ This study was approved by our institutional review board and adheres to Strengthening the reporting of observational studies in epidemiology (STROBE) guidelines. Waiver of informed consent was granted because of the retrospective nature of this study.

Volumetric Measurement

The EOR and RTV were calculated by volumetric measurement by 2 board-certified neuroradiologists with experience in oncological neuroradiology. The imaging analysis consisted of the volumetric measurement of 3 tumor components: enhancing tumor, necrosis, and nonenhancing fluid-attenuated inversion recovery (FLAIR) hyperintensity. The enhancing tumor volume and necrosis volume were obtained from standard-of-care MRIs on routine postcontrast 2-dimensional or 3-dimensional T1-weighted images before and after the first resection. FLAIR signal abnormality volume was measured on standard 2-dimensional or 3-dimensional T2-FLAIR images. Volumes were calculated using semiautomated software (TeraRecon Aquarius iNtuition viewer). By using the threshold tool, the software program selects volumes with a specific range of signal intensity values; then, the tumor boundaries are delineated using the region-of-interest tool. The software program automatically calculated the volume using computational algorithms.²²

Targeted Sequencing

All tumor samples were analyzed by targeted NGS using formalin-fixed paraffin-embedded tissues. Samples were sequenced in a Clinical Laboratory Improvement Amendments–certified laboratory (Foundation Medicine Inc.) using an assay that consistently targeted 205 tumor-related genes, including IDH1 and IDH2 and 26 gene rearrangements, as previously described.^21,23-25

Statistical Analysis

Descriptive statistics (N [%], mean [SD], or median [IQR]) were used to summarize patient characteristics. The Kaplan–Meier method was performed to estimate survival statistics and the difference between survival functions was tested by the 2-sided log-rank test. Multivariable Cox proportional hazards regression analyses were applied to assess the main effects of EOR or RTV on survival after controlling covariates, as well as each 2-way interaction. In addition, recursive partitioning analysis (RPA), as a nonparametric method, was performed to identify the optimal cutoff of EOR or RTV based on survival outcomes and determine the risk groups using classification and regression trees as the partitioning technique. Statistical analyses were performed using R (R Package Version 3.6.2). The P-values were 2-sided and considered statistically significant at P < .05.

In the analysis of the association between genetic alterations and EOR/RTV, we focused on genes or signaling pathway gene groups with ≥20 patients in both altered and wildtype groups, given that this will provide reliable statistical analysis. The TP53 pathway genes included TP53, MDM2, and MDM4. Retinoblastoma pathway genes included RB1, CDKN2A, CDKN2B, CDK4, and CDK6.²⁶ In each of these pathways, if any genes were altered, patients were interpreted as altered.

RESULTS

We identified 271 GBM IDH-wildtype cases with NGS data. Among them, 138 patients with relevant MRI and genetic data met our inclusion criteria. The median age of patients was 61 years (range: 13-87). EOR and RTV medians were 98.3% (range: 25.7%-100%) and 0.5 cm³ (range: 0-23.8 cm³), respectively. One hundred twenty-nine patients (93.5%) received radiotherapy and 129 (93.5%) were treated with temozolomide. One hundred twenty-six patients (91.3%) received combined adjuvant radiation and temozolomide (Table 1). Median progression-free survival was 8.28 months (IQR: 6.41-9.82), and median OS was 18.73 months (IQR 16.60-23.00). The median follow-up was 16.8 months (range: 1-83.2 months). We performed multiple models of RPA in combination with Cox multivariable analysis to identify the “ideal cutoff” for EOR and RTV in our patient population. In the 138 patients included in our study, patients with EOR ≥88% experienced 44% prolonged overall survival (OS) in multivariable analysis (hazard ratio: 0.56, P = .030). After using RPA combined with Cox multivariable analysis for EOR, we identified that the model that best identified survival was a combination of age and EOR in which patients older than 67.5 years and with an EOR <96% had worse survival compared with younger patients and/or patients with higher EOR after adjustment of multiple variables (hazard ratio [HR] 5.85, 95% confidence interval [CI] 2.89-11.91, P < .001), Table 2. Meanwhile, the best model that identified the best cutoff for RTV was a model including age and RTV, in which patients older than 67.5 years and with a RTV ≥1.09 cm³ had worse survival compared with younger patients and/or patients with decreased EOR after adjustment of multiple variables (HR 5.61, 95% CI 2.76-11.40, P < .001), Table 2.

TABLE 1.

Demographic, Clinical, and Genetic Variables of the Cohort

Variables	N = 138
Age at diagnosis, years, median (IQR)	61 (52, 69)
Male, N (%)	81 (58.7)
Race, N (%)
White	92 (66.7)
Hispanic	22 (15.9)
African American	16 (11.6)
Asian	6 (4.3)
Other	2 (1.4)
Preoperative KPS, N (%)
40 or <40	1 (0.7)
50	5 (3.6)
60	23 (16.7)
70	38 (27.5)
80	34 (24.6)
90	32 (23.2)
100	4 (2.9)
N/A	1 (0.7)
RTV, median (IQR) in cm3	0.5 (0, 2.5)
EOR, median (IQR) in %	98.3 (92.4, 100)
Radiation therapy, N (%)	130 (94.2)
Temozolomide, N (%)	129 (93.5)
Documented progression, N (%)	101 (73.2)
RB pathway mutations, N (%)	111 (80.4)
TERTp mutations, N (%)a	101 (80.2)
TP53 pathway mutations, N (%)	67 (48.6)
PTEN mutations, N (%)	62 (44.9)
EGFR mutations, N (%)	61 (44.2)
PIK3 pathway mutations, N (%)	59 (42.8)
NF1 mutations, N (%)	26 (18.8)
PDGFRA mutations, N (%)	21 (15.2)

EOR, extent of resection; KPS, Karnofsky performance status; N, number; N/A, not available; RB, retinoblastoma; RTV, residual tumor volume.

^aTERTp mutation information was not available for 12 patients; therefore, statistics were calculated with 126 patients.

TABLE 2.

The Threshold and Impact of RTV and EOR on Overall Survival Using RPA and Cox Proportional HR

Overall survival	RPA risk groups
	HR	95% CI	P value
EOR (%)
Age <67.5 + male	1.00	—	—
Age <67.5 + female	1.79	1.08-2.97	.024
Age ≥67.5 + EOR ≥96% + preop KPS ≥80	1.11	0.47-2.65	.810
Age ≥67.5 + EOR ≥96% + preop KPS <80	2.75	1.40-5.40	.003
Age ≥67.5 + EOR <96%	5.85	2.89-11.91	< .001
RTV (cm3)
Age <67.5 + male	1.00	—	—
Age <67.5 + female	1.79	1.08-2.97	.024
Age ≥67.5 + RTV <1.09 cm3 + preop KPS ≥80	1.11	0.47-2.65	.810
Age ≥67.5 + RTV <1.09 cm3 + preop KPS <80	2.81	1.43-2.65	.003
Age ≥67.5 + RTV ≥1.09 cm3	5.61	2.76-11.40	< .001

EOR, extent of resection; HR, hazard ratio; KPS, Karnofsky performance status; RPA, recursive partitioning analysis; RTV, residual tumor volume.

Statistically significant values are in bold.

The genetic evaluation of GBM IDH-wildtype revealed the following altered genes and pathways in ≥20 patients: retinoblastoma pathway (111/138, 80.4%), TERT promoter (TERTp) (101/126, 80.2%), TP53 pathway (67/138, 48.6%), PTEN (62/138, 44.9%), EGFR (61/138, 44.2%), PIK3 pathway mutations (59/138, 42.8%), NF1 (26/138, 18.8%), and PDGFRA (21/138, 15.2%), Table 1.

RPA identified that TP53 pathway alterations affected the association between EOR and survival (Figure 1A). Patients with alterations in the TP53 pathway and EOR <89% showed reduced OS, compared with TP53 pathway altered patients and EOR ≥89% (10.5 vs 18.8 months; HR: 2.78, 95% CI 1.25-6.21, P = .013, Figure 1B and Table 3). To note, in this multivariable model, radiotherapy was associated with improved survival (HR 0.22, 95% CI 0.07-0.72, P = .012) and increasing age with worse survival (HR 1.02, 95% CI 1.00-1.04, P = .015). RTV analysis did not show any difference between TP53 pathway-altered and TP53 pathway-wildtype patients. In patients without alterations in the TP53 pathway, EOR/RTV was not associated with OS.

FIGURE 1.

TP53 pathway relationship with overall survival. A, The recursive partitioning analysis. B, The Kaplan–Meier analysis using the log-rank test is shown in which TP53 pathway wildtype patients had the best survival with a median survival of 19.2 months, patients with TP53 pathway mutations ≥88.7% had an intermediated survival with a median of 18.8 months, while patients with TP53 pathway mutations with an EOR of <88.7% had the worse survival with a median overall survival of 10.5 months. EOR, extent of resection; OS, overall survival.

TABLE 3.

Thresholds and Impact of EORs by TP53 Pathway Status on Overall Survival Using Multivariable Cox Proportional Hazard Model

Overall survival	Multivariable analysis
	HR	95% CI	P value
EOR and PTEN status
TP53-mutant + EOR ≥89%	1.00	—	—
TP53-mutant + EOR <89%	2.78	1.25-6.21	.013
TP53 WT	0.71	0.45-1.12	.139
Age	1.02	1.00-1.04	.015
Sex
Female	1.00	—	—
Male	0.66	0.42-1.04	.075
Race/ethnicity
White	1.00	—	—
Hispanics	0.95	0.49-1.85	.890
African American	0.80	0.38-1.66	.541
Asian	0.62	0.26-1.51	.297
Others	1.86	0.42-8.14	.411
Preop KPS	1.00	0.98-1.02	.748
Radiation
No	1.00	—	—
Yes	0.22	0.07-0.72	.012
Temozolomide
No	1.00	—	—
Yes	0.48	0.17-1.40	.178

EOR, extent of resection; HR, hazard ratio; KPS, Karnofsky performance status.

Statistically significant values are in bold.

Furthermore, the RPA showed that by using PTEN alterations, IDH-wildtype GBMs can be stratified into different groups using both EOR and PTEN alterations status (Figure 2A). Patients with EOR >88% have the best OS. Meanwhile, among patients with EOR <87.69%, PTEN-altered patients had significantly worse survival than PTEN-wildtype (9.5 vs 15.4 months, P = .018) (Figure 2B). However, no relationship between RTV and PTEN was observed among our cohort. Moreover, the OS differences between PTEN-wildtype and PTEN-mutant patients in patients with EOR <88 remained significant after adjustment for age, sex, race, preoperative KPS, radiotherapy, and temozolomide in a multivariable Cox proportional hazard model (HR 4.53, 95% CI 1.91-10.71, P = .001, Table 4). To note in this multivariable model, radiotherapy was associated with improved survival (HR 0.18, 95% CI 0.05-0.60, P = .005), as well as male sex (HR 0.62, 95% CI 0.39-0.97, P = .038) and increasing age with worse survival (HR 1.02, 95% CI 1.00-1.03, P = .032).

FIGURE 2.

PTEN mutations' relationship with overall survival. A, The recursive partitioning analysis. B, The Kaplan–Meier analysis using the log-rank test is shown in which patients with an EOR >87.69% had the best survival regardless of their mutational status with a median survival of 20.4 months, patients with PTEN wildtype and EOR <87.69% had an intermediate survival with a median of 15.4 months, while PTEN mutant patients with EOR <87.69% had the worse survival with a median overall survival of 9.5 months. EOR, extent of resection; OS, overall survival.

TABLE 4.

Thresholds and Impact of EORs by PTEN Status on Overall Survival Using Multivariable Cox Proportional Hazard Model

Overall survival	Multivariable analysis
	HR	95% CI	P value
EOR and PTEN status
EOR ≥87.7	1.00	—	—
PTEN WT + EOR <88%	1.36	0.73-2.55	.338
PTEN-mutant + EOR <88%	4.53	1.91-10.71	.001
Age	1.02	1.00-1.03	.032
Sex
Female	1.00	—	—
Male	0.62	0.39-0.97	.038
Race/ethnicity
White	1.00	—	—
Hispanics	1.16	0.60-2.23	.666
African American	0.82	0.39-1.69	.586
Asian	0.79	0.33-1.88	.597
Others	2.36	0.55-10.13	.247
Preop KPS	1.00	0.98-1.02	.743
Radiation
No	1.00	—	—
Yes	0.18	0.05-0.60	.005
Temozolomide
No	1.00	—	—
Yes	0.57	0.19-1.68	.307

EOR, extent of resection; HR, hazard ratio; KPS, Karnofsky performance status.

Statistically significant values are in bold.

The other genes/pathway alterations evaluated (EGFR, retinoblastoma pathway, NF1, PDGFRA, and PIK3CA) did not appear to influence the association of EOR/RTV and survival in this cohort.

DISCUSSION

In this study, we demonstrate the relationship between the EOR, genetic alterations, and survival in a cohort of GBM IDH-wildtype patients. Our results show that 3 groups can be distinguished using EOR and PTEN alterations. Group 1: patients with EOR >87.69 with the best OS ∼20.4 months, group 2: PTEN-wildtype patients with EOR <87.69 with an intermediate OS ∼ 15.4 months, and finally group 3: PTEN-mutant patient with EOR <87.69 with the worst OS ∼ 9.5 months (P < .001). In addition, within the TP53 pathway (TP53, MDM2, and MDM4 alterations), EOR is remarkably relevant among patients with genetic alteration, while its importance is not so clear in patients without alterations in the aforementioned pathway.

EOR has been studied by multiple groups that have identified different thresholds (eg, 70%,⁹ 78%,⁸ 98%,⁷ and 100%³), which appear to decrease as larger data sets are studied. In our study, the threshold of survival was determined by ∼88% EOR. However, as pointed out by Solheim et al,²⁷ dichotomizing survival by a threshold might not be as clinically relevant for patients because clinically, but not statistically, significant survival benefit could be lower than our threshold and this might just reflect the current threshold that statistically reflects survival with the current data set. Therefore, as pointed out by other groups,^6,28 we recommend maximal safe resection in all patients with GBM. In addition, in our patient population, a RTV of ≥1.09 cm³ was strongly correlated with worse survival. In some studies, RTV has been found to be a better prognostic factor for survival in GBM from the oncological standpoint^11,12 and has a better inter-rater agreement.¹¹ Therefore, the recent Response Assessment in Neuro-Oncology (RANO) resect effort stratifies patients solely on the RTV.

As GBM has been further studied, genetic biomarkers such as IDH1/IDH2 and 06-methylguanine-DNA methyltransferase (MGMT) have been identified to be key to patient survival.^14,29,30 Although other genetic markers in IDH-wildtype GBM have been proposed,^25,31,32 more studies are needed to elucidate their role in the outcome of patients with GBM. Therefore, EOR survival benefits among patients diagnosed with GBM IDH-wildtype with different genetic makeups might differ. This has been shown by our group in a recent study of reoperation in which some patients with GBM benefit the most from resection depending on the genetics of their tumor.³³

In newly diagnosed GBM IDH-wildtype, different EOR thresholds associated with survival have been observed among patients with MGMT methylation compared with nonmethylated cases.¹⁹ In this study,¹⁹ it was hypothesized that unmethylated tumors might have a lower EOR threshold for survival improvement as temozolomide chemotherapy does not improve survival as seen in methylated tumors.²⁹ Other studies evaluating the EOR threshold in molecular subtypes of GBM evaluated the EOR threshold among different methylation subclasses of GBM.³⁴ However, there are no studies evaluating EOR thresholds among different genetic subtypes of GBM.

PTEN alterations and their relationship with survival in GBM are controversial.^35,36 However, recent studies have shown that PTEN status correlates with chemotherapy response,³⁷ TTFields,³⁸ or radiotherapy.^39,40 It is conceivable that GBM IDH-wildtype with PTEN alterations with a lower EOR (88% in this study) has worse survival. However, this will need to be further investigated by other groups to validate our results.

TP53 alterations in GBM IDH-wildtype have been associated with primitive neuronal component histology, a subtype that has poor survival.³⁶ Interestingly, our previous study evaluating reoperation for GBM IDH-WT³³ identified that reoperation is more beneficial in patients harboring TP53 mutations. Congruent with these results in the reoperation setting, the current study identified that EOR was crucial in patients diagnosed with GBM IDH-wildtype harboring TP53 pathway alterations, but not in those without alterations in the pathway. It is unclear whether this results from a more aggressive tumor behavior in patients with alterations in this pathway, or whether other therapeutic avenues are more effective in patients without alterations. Further investigation with larger groups is warranted.

Our study did not identify other genes such as EGFR, CDKN2A/B, and NF1 (previously associated with better outcomes after reoperations) as influencing the association of EOR/RTV and survival in this cohort.³³

Neurosurgeons should strive toward maximal safe resection in all patients with GBM as supratotal resection has been shown to improve survival even when compared with gross total resection.^11,18 However, postoperative neurological complications should be avoided because they hamper improved outcomes⁴¹ and quality of life. Our findings suggest that the EOR benefits are not homogeneous among different GBM IDH-wildtype subtypes. Patients harboring PTEN or TP53 pathway alterations benefit more from aggressive resections. This could potentially suggest that neurosurgeons may consider reoperation to optimize an initial suboptimal resection in a subset of patients with favorable genetic status.⁴² In addition, as new technologies for accurate preoperative diagnosis (eg, plasma or cerebrospinal fluid biomarkers⁴³) become more accurate and accessible, as well as better intraoperative rapid diagnosis tools such as stimulated Raman histology⁴⁴ become available, a more informed decision of the surgical resection aggressiveness needed to improve survival according to tumors genetic makeup could be performed, providing precision-guided surgical resections according to individual tumor biology.

Limitations

The limitations of our study include its retrospective design and potential selection bias because not all patients with newly diagnosed GBM in our institution underwent NGS. Also, supratotal resection could not be evaluated in this study because evaluating multiple genetic alterations in smaller EOR subgroups would not provide reliable statistical analysis, thus multi-institutional collaborations to evaluate the impact of genetic mutations other than MGMT and IDH in glioblastoma are needed. In addition, not all patients underwent MRI within 48 to 72 hours, which would allow us to assess EOR. EOR could not be consistently evaluated in some molecular subgroups, given its relatively small number. MGMT promoter status was unavailable in most patients. Given the relatively small sample size, a lack of findings in some groups may be due to a lack of statistical power. Finally, in recent studies, RTV has been demonstrated as a prognostic factor and more important than EOR. Therefore, we evaluated RTV in the context of different genetic alterations, but we did not identify a strong correlation among them. Because EOR is calculated based on the preoperative tumor volume, patients with the same EOR can have different RTVs. Moreover, RTV evaluation is less demanding and more homogeneous across different cancer centers.¹¹

CONCLUSION

Our results suggest that a subset of molecularly defined GBMs IDH-wildtype may benefit more from aggressive resections. Re-resections to optimize EOR at initial diagnosis might be beneficial in a subset of molecularly defined GBMs. Molecular alterations should be taken into consideration for surgical treatment decisions in GBM IDH-wildtype.

Funding

The research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number K08CA241651 (LYB). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosures

The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.