De ning the Impact of Adjuvant Treatment on the Prognosis of Patients with Inoperable Glioblastoma Undergoing Biopsy-only: Does the Survival Bene t Outweigh the Treatment Effort?


 Purpose: Patients with inoperable glioblastoma (GBM) usually experience worse prognosis compared to those in whom gross total resection (GTR) is achievable. Considering the treatment duration and its side effects identification of patients with survival benefit from treatment is essential to guarantee the best achievable quality of life. The aim of this study was to evaluate the survival benefit from radio-chemotherapy and to identify clinical, molecular, and imaging parameters associated with better outcome in patients with biopsied GBMs. Methods: Consecutive patients with inoperable GBM, who underwent tumor biopsy at our department from 2005 to 2019 were retrospectively analyzed. All patients had histologically confirmed GBM and were followed up until death. The overall survival (OS) was calculated from date of diagnosis to date of death. Clinical, radiological and molecular predictors of OS were evaluated. Results: A total of 95 patients with biopsied primary GBM were enrolled in the study. The mean age was 64.3±13.2 years, 56.8 % (54/95) were male and 43.2 % (41/95) female. Mean OS in the entire cohort was 9 months. After stratification for adjuvant treatment a higher median OS was found in the group with adjuvant treatment (7 months, range 2-88) compared to the group without treatment (1 month, range 1-5) Log-rank test, p<0.0001.Conclusion: Patients with inoperable GBM undergoing biopsy indeed experience a very limited OS. Adjuvant treatment is associated with significantly longer OS compared to patients not receiving treatment and should be considered, especially in younger patients with good clinical condition at presentation.


Introduction
Glioblastoma multiforme (GBM) is associated with very limited survival odds requiring a well-balanced management between aggressive treatment to increase the chance of longer survival and the preservation of quality of life. The median overall survival (OS) of GBM patients is 15 months, whereas only 3-5% survive longer than 3 years, who are then already considered long-term survivors. The 10-year survival rate with GBM was estimated to be 0.71% [21]. Besides younger age and good clinical status at diagnosis with independency during every-day life activities, the extent of resection (EOR) is considered as clinically relevant prognostic factor in GBM patients [10]. Since application of 5-aminolevulin acid (ALA) has signi cantly improved the rates of gross total resection (GTR) resulting in increased over all (OS)-and progression free survival (PFS), GTR whenever achievable has been pursued as a surgical goal during glioma surgery [18]. However, the risk of postoperative neurological de cits increases with radical tumor resection outlining a delicate line between reaching GTR and avoiding neurological de cits.
Especially in eloquently located tumors GTR is often not feasible, even by means of intraoperative neuromonitoring tools [9]. Since neurological de cits are directly associated with worse outcome, the bene t of more radical tumor resection at the expense of neurological de cits is questionable.
Considering the increasing knowledge about GBM growth-patterns as widespread in ltrating tumor at the time of diagnosis, GTR does not seem to be a reasonable surgical goal in patients with multifocal or eloquent GBM [12,15,20]. Hence, most neurooncological centers only perform a tumor biopsy for histological con rmation and treatment planning in these patients. The patient population with inoperable GBM is expected to have a very limited survival since they cannot receive maximal available treatment. The actual survival bene t of the radio-chemotherapy in this patient population has not been determined yet. Considering the limited prognosis of these patients on the one side and the strain of adjuvant treatment on the other side, it is of great clinical interest to evaluate the impact of radiochemotherapy on survival in this speci c patient population. The objectives of this study were: 1-to shed light on this heterogenous patient population with solely biopsied GBM and to elucidate the criteria for inoperability, 2-to assess whether the survival bene t from tumor treatment overweighs the associated burden of radio-chemotherapy regarding the very limited life expectancy, 3-to identify radiological and molecular tumor-speci c factors alongside with patient-speci c parameters associated with longer OS in patients with biopsied GBM considered inoperable.

Patient population
We performed a retrospective analysis of a consecutive and homogenous patient cohort from the Stuppera. GBM was histologically con rmed after performing a biopsy of tumors considered inoperable. The reasons for considering the tumor as inoperable were extracted from medical records and operations reports. To maintain comparable standards of treatment, patients treated at our department in the time interval from 2005 to 2019 were included in the analysis of the study. Patient-speci c parameters like age at diagnosis, Karnofsky Performance Status (KPS) at presentation, tumor-speci c parameters including tumor location and distribution on imaging, as well as molecular markers such as the presence of MGMT (O-6-methylguanine-DNA methyltransferase) promotor methylation and IDH (isocitrate dehydrogenase)mutation were documented. Additionally, data regarding adjuvant treatment and treatment response as seen on imaging during the follow-up examinations were gathered. Treatment decisions were made after interdisciplinary case discussion by the institutional tumor-board for tumors of the central nervous system. Volumetric analysis of the tumor on initial magnetic resonance imaging (MRI) was performed using the Brainlab software Elements (Brainlab® Munich, Germany).
Outcome parameters OS was de ned as the time in months from diagnosis to the date of death. Data on patients who were lost to follow-up, were censored. The impact of patient-speci c, radiological and molecular tumor-speci c markers on survival were evaluated. The treatment response on imaging was analysed according to the response assessment in neuro-oncology criteria (RANO), at 3-and 6-months follow-up imaging. PFS was de ned as time from diagnosis to tumor progression on follow-up imaging according to the RANO criteria for tumor progression [24].

Statistical analysis
The statistical analyses were performed by means of GraphPad Prism (Version 9, GraphPad Software, San Diego, CA, USA). For the presentation of baseline data, descriptive statistics and frequency, distribution analysis was done. Continuous variables are depicted as means ± standard deviation (SD), categorical variables as frequencies or percentages. To analyze the difference of continuous variables and categorical data, t-and Chi-square tests were applied respectively. By performing univariate and multivariate linear regression models we analysed the effects of patient-/tumor characteristics on OS. Statistical signi cance was set as p < 0.05. Overall survival rates were calculated by Kaplan Meier method applying a Log-rank test.

Patient characteristics
A total of 518 patients with newly diagnosed GBM were treated at our department between 2005 and 2019 (15 years), of whom 95 patients (18.3%) received a tumor biopsy without subsequent resection and were eligible for inclusion into further analysis of this study. Mean age was 64.3±13.2 (range 29 to 88) years, 56.8% (54/95) were male and 43.2% (41/95) were female. The median KPS at admission was 70% (range 40-100). Baseline characteristics are depicted in Table 1. The mean age of patients who received adjuvant treatment was signi cantly lower compared to the patient group without tumor treatment, while the other baseline characteristics were not signi cantly different.

Tumor characteristics
The decision-making criteria whether a tumor was considered inoperable were based on its localization and distribution on initial imaging. Main reasons for a biopsy-only approach were expected neurological de cits by performing GTR. For example, in tumors involving eloquent brain structure or the corpus callosum (bilateral in ltration=butter y GBM), basal ganglia, brain stem or primary motor cortex, GTR was not reasonably achievable, hence, biopsy-only was performed. Multifocal tumor growth was another common criterium for biopsy-only instead of a resection of all tumor loci. Examples of different localizations and distributions on initial imaging are given in Fig1.

Discussion
The primary objective of this study was to identify factors associated with longer survival in a patient population with newly diagnosed inoperable GBM with or without subsequent tumor-speci c treatment after histological diagnosis con rmation by biopsy-only. The focus of the analysis was put on the evaluation of actual survival bene t from tumor treatment in this very speci c subgroup in relation to the therapy burden. Hence, our long-term aim is to avoid possibly futile, however straining, therapy in this patient group with very limited survival odds. In 18% of our consecutive GBM patient cohort, the tumor was considered inoperable due to tumor growth patterns affecting key brain structures and therefore not permitting GTR without causing unacceptable neurological de cits. To the best of our knowledge, this is the largest patient series with newly diagnosed inoperable GBM reported in the literature so far. Our study con rmed a very limited OS (one-month median OS) in patients with inoperable GBM without tumorspeci c treatment. In contrast to this, the patient cohort with inoperable GBM, who received tumor treatment (radio-chemotherapy) after histological con rmation of the diagnosis by biopsy, exhibited a signi cantly longer OS, clearly demonstrating a survival bene t from treatment in this patient population.
However, the median OS of patients with inoperable GBM was still signi cantly lower compared to the reported median OS of operable GBM (9 vs. 15 months), highlighting the prognostic role of GTR in GBM patients. The identi cation of further clinical, radiological and molecular parameters associated with prolonged survival is of great clinical relevance to facilitate targeted treatment in patients with GBM, in whom GTR is not reasonably achievable.

De nition of tumor operability and prognostic role of extent of resection
The criteria applied for de ning inoperability are still a matter of discussion, even though a neurosurgical expert consensus concerning this de nition exists. Several factors can affect the perception of tumor resectability such as the experience of the neurosurgeon as well as the technical facilities of the institution [8,14]. In our patient series, tumor localization within eloquent regions and multifocal tumor distribution on initial MRI were the main determining criteria for inoperability. Nevertheless, advances in functional imaging over time might have had facilitated a better interpretation of tumor relation to eloquent regions. Furthermore, surgery can not only contribute to neurooncological bene t but can also result in better clinical condition by prevention of threatening neurological de cits. Thus, resection of a large space-occupying mass or resection of a symptomatic lesion to relieve symptoms in a primary multifocal tumor may be discussed individually. A longitudinal study of patient-reported quality of life, GTR was not only associated with higher PFS and OS but also with higher quality of survival [16]. On the other side there is a risk for surgery-related complications including neurological de cits with a subsequent negative impact on patients' clinical condition. A meticulous weighting up of operative risks and expected bene ts after diagnosis is crucial in order to provide the best achievable survival advantage from surgical treatment without additional impairment of patients' performance status. Even though several studies have shown a survival advantage through cytoreductive surgery in GBM patients there is an ongoing controversial discussion regarding the threshold of extent of resection (EOR) ranging from 78% to 100% [14,17]. While it is a common practice to perform GBM-resection only if GTR seems achievable, some authors hold the opinion that no threshold can be set, below which tumor resection does not extent survival. They point out the need of balancing between clinically futile and clinically worthwhile extent of tumor resections rather than a strict cutoff line [4]. After performing a systematic review of the literature and a meta-analysis Brown et al. reported a low to moderate evidence that GTR substantially improves OS and PFS compared to subtotal resection [4]. Therefore, a trade-off between the expected survival bene t and the quality of life is necessary, since most GBM patients would rather choose better quality of life than longer life with poor quality. The same applies to tumor treatment with radio-chemotherapy. The bene t of tumor treatment in patients with inoperable GBMs has not been determined yet. Radiation and/or chemotherapy with varying regimens have been already evaluated in patient cohorts with unresectable GBMs showing a survival bene t [2,3]. These reports are in line with our study, where tumor treatment was associated with signi cantly longer median OS compared to patients, without tumor therapy. However, there are possible confounders that might have impacted this nding: patients receiving radio-chemotherapy were on average younger, which per se positively in uences survival. This may have resulted in a natural selection bias in our study due to patients who did not elect to be treated after a diagnosis of GBM. It is possible that these patients in the no-treatment group tended to decompensate rapidly after diagnosis, had a lower KPS upfront, or had other signi cant comorbidities that in uenced their decision. Therefore, patients in the no-treatment group would experience decreased survival time due to their intrinsic worse clinical condition. Nevertheless, multivariate analysis revealed the presence of IDH-mutation and completed radio-chemotherapy according to Stupp protocol as independent prognostic factors, whereas age was no independent predictor of survival. The identi cation of further tumor-speci c factors in uencing OS would facilitate better treatment planning and counseling of patients with inoperable tumors.

Tumor-speci c predictors of survival in patients with inoperable GBM
Over the last years several radiological and molecular factors were evaluated for their prognostic impact in GBM patients. Different radiological tumor-characteristics including initial tumor volume, growth patterns, contact to glioma stem-cell-zones, ventricle contact and involvement of the corpus callosum were evaluated as prognostic factors in our study. None of the tumor-speci c locations correlated directly with OS in our patient cohort. However, the most common tumor location in the no-treatment group in our study involved the CC (80%). As it is known that patients with butter y GBM might experience severe personality changes this speci c location might have in uenced the capacity to tolerate adjuvant treatment. The topic of tumor-speci c locations is under discussion in the current literature as different tumor locations offer or limit GTR and consequently in uence OS. Recently the team around Müller et al.
tried to grasp this topic by introducing the "expected residual tumor volume" (eRV) and the "expected resectability index" (eRI), hereby quantifying treatment decisions, resectability and survival [13]. The authors demonstrated that the higher the (expected) residual tumor volume and the less the (expected) resectability, the shorter the expected OS. However, the authors acknowledged that different tumor locations might result from different molecular subtypes and therefore might have had a different natural course of progression. So far, four clinically relevant GBM subtypes were identi ed, each with a distinct molecular pattern, response to standard therapy and natural course [22]. Until now, no association between localization and tumor subtype has been determined and therefore, further strati cation for tumor subtype and tumor treatment might have been a valuable addition. The presence of IDH-mutation and MGMT promotor methylation are well-established molecular markers associated with a favorable outcome [4,25]. In our study, only the presence of IDH-mutation was independently associated with longer OS.

Treatment response in patients with inoperable GBM
During follow-up of our cohort, we evaluated treatment response according to the RANO criteria [24]. Most patients who underwent tumor biopsy presented with disease progression within 6 months, but the patient group with tumor treatment demonstrated a prolonged PFS. To date, several studies proved a survival bene t for GBM undergoing GTR versus GBM with subtotal resection or biopsy [1,[5][6][7]. However, the group of patients with biopsied GBM was mostly not strati ed for tumor treatment. A study by Balaña et al. from 2007 evaluated the survival of "biopsy-only GBM" and revealed comparable results to our study. Their cohort comprised 34 patients with grade III and IV gliomas and was therefore more heterogenous than our cohort. They also found a signi cant survival bene t of patients undergoing biopsy followed by radio-chemotherapy treatment compared to those without any further therapy [1]. In both studies about 60-70% of patients received tumor treatment, which is very similar to the nding of our study (66%). Interestingly all treatment regimens led to prolonged OS compared to the patients without further treatment. The reasons why some patients did not receive any tumor treatment were a reduced KPS, death before therapy initiation or due to patients' or their family's refusal [1]. While individualized therapy concepts are of increasing importance and advancements are made regarding biomarkers and more sophisticated therapies in GBM, most treatment strategies are still based on KPS, age and MGMTpromotor status. At present, there are multiple studies that attempt to nd individualized therapies according to the molecular subtype [23] that hold the potential for future effective GBM therapy.

Limitations of the study
There are several limitations due to the retrospective nature of the study, that need to be acknowledged. Due to missing data on molecular markers in the patients biopsied before 2016, our study could not provide a more diverse evaluation of the molecular markers in uencing OS. Also, complete radiological data were only available in 24 of 95 patients. Therefore, the results have to be interpreted with caution due to the small sample size. Furthermore, no data on quality of life was available, hence, preventing an evaluation of the compromise of quality of life during radio-chemotherapy, which is a relevant issue and represents a limitation of the study.

Conclusion
Patients with inoperable GBM undergoing biopsy indeed experience a very limited OS. Tumor treatment is associated with signi cantly longer OS compared to patients not receiving treatment and should be considered, especially in younger patients with good clinical condition at presentation. However, it should be kept in mind that despite bene t of tumor treatment the patient population with inoperable GBM still has a very limited median OS of 9 months with a tumor treatment duration of at least 6 months.

Funding
The study was not funded. Overview of the different tumor localizations in our patient cohort with different growth patterns on initial contrast enhanced MRI: GBM with bilateral in ltration growing as a butter y GBM involving the corpus callosum (upper left), multifocal GBM with multiple contrast-enhanced tumor areas (upper right), motor cortex GBM involving the precentral gyrus (lower left), and GBM within the basal ganglia (lower right).

Figure 2
Examples of good and poor response to tumor treatment as seen on contrast enhanced MRI and based on the RANO (response assessment in neuro-oncology criteria) criteria, showing varying responses to treatment according to the RANO criteria comparing initial MRI with contrast with radiological response at follow-up at 3 and 6 months after tumor treatment: A: partial radiological response (RANO 2), B: stable disease (RANO 3) and C: disease progression as de ned by RANO 4.

Figure 3
A Kaplan-Meier curves showing signi cantly longer OS in the patient group that received tumor treatment compared to the patient group without tumor treatment (log-rank test, p<0.0001).
B Kaplan-Meier curves showing the longest OS within the group that received R-CH according to the Stupp protocol, followed by other treatment protocols including CeTeG or metronomic temozolomide and RT alone, with the shortest OS in the group without tumor treatment (log-rank test, p<0.0001).