Elderly Gliobastoma Patients: The Impact of Surgery and Adjuvant Treatments on Survival: A Single Institution Experience

Introduction. Elderly glioblastoma (GBM) patients often show limited response to treatment and poor outcome. Here, we provide a case series of elderly GBM patients from our Institution, in whom we assessed the clinical characteristics, feasibility of surgical resection, response to adjuvant treatments, and outcome, along with the impact of comorbidities and clinical status on survival. Patients and Methods. We included patients ≥ 65-year-old. We collected information about clinical and molecular features, extent of resection, adjuvant treatments, treatment-related complications, and outcome. Results. We included 135 patients. Median age was 71 years. In total, 127 patients (94.0%) had a Karnofsky Performance Status (KPS) ≥70 and 61/135 (45.2%) a Charlson Comorbidity Score (CCI) > 3. MGMTp methylation was found in 70/135 (51.9%). Subtotal resections (STRs), gross-total resections (GTRs), and biopsies were 102 (75.6%), 10 (7.4%) and 23 (17.0%), respectively. Median progression-free survival and overall survival (mOS) were 8.0 and 10.5 months for the whole cohort. Notably, GTR and radio-chemotherapy with temozolomide in patients with MGMTp methylation were associated with significantly longer mOS (32.8 and 44.8 months, respectively). In a multivariable analysis, risk of death was affected by STR vs. GTR (HR 2.8, p = 0.002), MGMTp methylation (HR 0.55, p = 0.007), and KPS at baseline ≥70 (HR 0.43, p = 0.031). Conversely, CCI and post-surgical complications were not significant. Conclusions. Elderly GBM patients often have a dismal prognosis. However, it is possible to identify a subgroup with favourable clinical and molecular features, who benefit from GTR and radio-chemotherapy with temozolomide. A comprehensive prognostic score is needed to guide treatment modality and predict the outcome.


Introduction
Glioblastoma (GBM) is the most frequent and aggressive primary brain tumour in adults, accounting for 14.5% of all brain tumours and 57.7% of gliomas [1]. The incidence of GBM increases with age. In fact, according to the Central Brain Tumour Registry of an American (CBTRUS) statistical report from the 2013-2017 period, the incidence of GBM is 3.23 per 100,000 people per year, reaching its peak in people 75-84-year-old (15.30 per 100,000 people per year) [1]. Moreover, as life expectancy is progressively increasing, the incidence of GBM in the elderly population will further rise in the future.
In this paper, we provide a case series of elderly patients with GBM from our Institution, in whom we assessed the clinical characteristics, feasibility and effectiveness of surgical resection, response to adjuvant treatments, and outcome, along with the impact of comorbidities and clinical status on survival. The aims of our study were: to investigate the impact of surgery and adjuvant treatment on survival; to assess whether comorbidities significantly impacted the choice of treatments and outcome; to assess how clinical characteristics and outcome were distributed across different classes of age; and to identify a subgroup of elderly patients with more favourable prognoses.

Retrospective Collection of Patients' Data
From 2015 to 2020, we collected in a specific database all elderly (≥65 years) GBM patients who have been diagnosed and treated at the Division of Neuro-Oncology, University and City of Health and Science Hospital of Turin.
The database collected all the information about clinical, pathological, molecular, and therapeutic characteristics of patients. From a dataset of 160 patients, we have selected only those patients who underwent a surgical procedure, with a histologically confirmed diagnosis of GBM IDH-wildtype. One hundred and thirty-five patients were available for the analysis.
The presence of IDH mutation was ruled out in all cases by Sanger gene sequencing. The MGMT promoter methylation status was assessed by polymerase chain reaction (PCR).
Extent of resection (EOR) was defined as gross-total or subtotal based on the presence of a contrast-enhancing residual tumour on MRI after 24-72 h from surgery. Response Assessment in Neuro-Oncology (RANO) criteria were used to evaluate MRI response to treatments [2]. Treatment-related adverse events were defined according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.

Statistical Analysis
Baseline characteristics of patients included in the analysis are summarised using median and interquartile range (IQR) and percentages and frequencies (n, %). We adopted age at surgery as a surrogate of age at diagnosis. The observation period for progressionfree survival (PFS) and overall survival (OS) started on the date of surgery until the date of recurrence or death, respectively, or until the last follow-up visit (censoring).
The distribution of characteristics between molecular subgroups were evaluated by the Mann-Whitney U test for continuous variables and the Chi-square test or Fisher's exact test for categorical variables. Kaplan-Meier curves were drawn for PFS and OS and a Cox proportional hazard model was employed to estimate the crude and the multivariableadjusted hazard ratios (HRs) with 95% confidence intervals (CIs) and to evaluate possible predictors of recurrence or survival. The proportional hazard assumption was also verified by graphical checks and formal tests based on Schoenfeld residuals.
The analysis was performed by IBM SPSS Statistics v.28 software.

Ethical Approval
The project was approved by the Review Board of our Institution. Informed consent to collect and analyse clinical and pathological/molecular data was obtained from all subjects who were alive at the time of start of the study or from relatives in case of death of the subjects, according to ethic regulations for retrospective studies of the local Ethics Committee.

Complications after Surgery and Correlation with the Extent of Resection (EOR)
Fifty-one (37.7%) patients presented clinical complications and/or neuroradiological impairment within one month of surgery before starting adjuvant treatments. In 28 cases, (20.7%) clinical complications occurred in the first week after surgery (see Table 2 for detailed information).
The occurrence of clinical complications after surgery was not associated with CCI or KPS at baseline. Moreover, there was no association between the entity of resection and occurrence of clinical complications, which was accounted for in 8/23 (34.8%) patients after biopsy, for 27/102 (26.5%) after subtotal resection, and for 1/10 (10.0%) patients after gross-total resection (p = 0.333).

Clinical Characteristics across Different Classes of Age
We evaluated the distribution of different clinical factors across different classes of age of our cohort (65-69; 70-74; 75-79; ≥80 years). First, older patients had a higher CCI (CCI score > 3). As age influences the final CCI score, we subtracted the value of age from the total CCI score to consider the weight of comorbidities only; even in this case, people older than 80 had a significantly higher index, being age-adjusted CCI > 3 in 3/52 (5.8%), 1/51 (2.0%), 0/23 and 2/9 (22.2%) among patients 65-69, 70-74, 75-79, and ≥80-year-old, respectively (p = 0.033). Higher age was also associated with higher incidence of clinical complications after surgery. Conversely, other clinical factors (such as incidence of seizures and multifocal presentation) as well as prevalence of MGMTp methylation did not differ among age subgroups. Regarding adjuvant treatments, 6-week concomitant RT/TMZ followed by adjuvant TMZ prevailed among patients younger than 75 years (31/103, 30.1% vs. 2/32, 6.2%), whereas those older than 75 years underwent more frequent 3-week RT followed by adjuvant TMZ (5/32, 15.6% vs. 8/103, 7.8%) (see Table 3). MGMT, O(6)-methylguanyl DNA methyltransferase promoter; mOS, median overall survival; mPFS, median progression-free survival; RT, radiotherapy; TMZ, temozolomide. * CCI was modified by subtracting the value attributed to age from the total score, to make comorbidities the only factors included in the index.

Outcome across Different Classes of Age and According to Clinical Performance Status and Comorbidities
Patients 65-69 years old had a mPFS of 7.7 months and a mOS of 11.0 months, whereas those 70-74 years old had a mPFS of 9.3 months and mOS of 10.1 months, and those 75-79 years old had a mPFS of 4.2 months and mOS of 8.8 months. However, differences in mPFS and mOS across different age classes were not significant (Table 3). Remarkably, the nine patients older than 80 years had quite a long mPFS (19.8 months) and mOS (28.7 months); this could be explained by the fact that all those nine patients received an active treatment after surgery, mostly RT/TMZ followed by adjuvant TMZ (3-week and 6-week schedule in 6/9 and 1/9, respectively), and upfront TMZ in two cases, which could represent a selection bias for a better outcome despite the advanced age.

Outcome According to Extent of Resection and Adjuvant Treatments
EOR displayed a significant impact on outcome, with mPFS and mOS being longer for patients undergoing gross-total resection (18.8 and 32.8 months, respectively) than subtotal resection (6.5 and 9.9 months) or biopsy (1.0 and 6.0 months) (p = 0.002 for mPFS and p < 0.001 for mOS models, respectively). Clinical complications after surgery did not significantly affect mPFS (8.2 vs. 7.9 months, p = 0.406) and mOS (10.7 vs. 10.5 months, p = 0.675). Among patients undergoing adjuvant treatments (111, 82.2%), those who underwent 6-week RT/TMZ followed by TMZ had the longest mPFS (18.8 months) and mOS (28.4 months), followed by those who underwent RT followed + TMZ (mPFS 12.7 months; mOS 23.3 months) and 3-week combined RT/TMZ + TMZ (mPFS 10.4 months; mOS 16.0 months). Ultimately, mPFS and mOS of patients who underwent RT alone after surgery were 3.9 and 6.0 months, respectively, whereas in patients who were treated with TMZ alone they were 3.3 and 8.3 months, respectively.
Among patients who underwent concomitant RT/TMZ + TMZ who displayed more frequent treatment-related adverse events (Table 5) Median progression-free and overall survival according to EOR and adjuvant treatment are reported in Table 7.

Outcome According to MGMTp Status within Groups of Different Treatment Modalities
Within the whole cohort, MGMTp methylation was associated with significantly better mPFS (11.7 vs. 6.0 months, p < 0.001) and mOS (19.0 vs. 9.8 months, p < 0.001).
We also evaluated mPFS and mOS of patients undergoing different adjuvant treatment modalities after surgery according to MGMTp status. The presence of MGMTp methylation correlated with a significantly longer mPFS and mOS among patients undergoing 6-week RT/TMZ + TMZ (mPFS: 20.8 vs. 9.6 months, p < 0.001; mOS: 44.8 vs. 10.7 months, p < 0.001) and 3-week RT/TMZ (mPFS: 19.5 vs. 7.6 months, p < 0.001; mOS: 27.1 vs. 12.9 months, p = 0.002). A non-significant trend for better mOS was also seen among patients with MGMTp methylation treated with upfront TMZ (8.3 vs. 4.6 months, p = 0.090); conversely, among patients undergoing RT alone or RT followed by TMZ, the presence of MGMTp methylation did not significantly affect the outcome (Table 8).

Multivariable Analysis of Prognostic Factors on Progression-Free and Overall Survival
In a multivariable analysis of the whole cohort, factors that significantly reduced the risk of progression in a multivariable analysis were: MGMTp methylation, large surgical resection, and adjuvant radio-chemotherapy; factors that significantly reduced the risk of death were: KPS at baseline ≥ 70, MGMTp methylation, large surgical resection, adjuvant radio-chemotherapy, and second line treatments (Table 9).  To conclude, the subgroup of patients who displayed the best outcome were those with KPS at baseline ≥ 70, with MGMTp methylation, who underwent gross-total resection and adjuvant radio-chemotherapy with TMZ: in this subgroup, mPFS was 35.5 months and mOS was not reached.

Discussion
Our single, large institutional study on elderly GBM confirms that glioblastoma in the elderly population (≥65 years) is a highly aggressive tumour, with mPFS and mOS being 8.0 months and 10.5 months, respectively. However, our data show that there is still some heterogeneity within this cohort of patients, giving rise to still unanswered issues.
First, which is the most appropriate threshold to define elderly age is not clear, as this population may be further stratified by different classes of age based on clinical characteristics and outcome. In fact, a non-significant (but still noteworthy) trend for worse survival in older patients was seen in our cohort, when comparing 65-69 year-old patients (mPFS 7.7 months; mOS 11.0 months) to 70-74 year-olds (mPFS 9.3 months; mOS 10.7 months), and to 75-79 year-olds (mPFS 4.2 months; mOS 8.8 months) ( Table 3). In our series, a higher prevalence of comorbidities and an increased risk of clinical complications after surgery were significantly associated with higher age, which explains the poorer outcome of the oldest patients within the series (Table 3).
Second, which is the most appropriate comorbidity score in elderly patients with GBM has not been established thus far. The Charlson Comorbidity Index was developed to predict the 10-year mortality of patients, scoring from 1 to 6 depending on the risk of death from comorbidities, and it is uncertain whether such a score may be adequate in glioblastoma patients with dramatically short survival [3]. In our series, the CCI was not a reliable predictor of outcome, except for the youngest patients of the cohort (65-69 years old), who displayed a poorer survival with CCI > 3 (Supplementary Table S2). This was probably due to the prominent weight of age rather than comorbidities in determining the CCI score of our patients. In fact, the number of patients without comorbidities was considerable (67/135, 49.6%); in these cases, age was the only factor determining the CCI score (Table 1).
A strong correlation between CCI and outcome was not found in previous studies involving elderly GBM patients. In a study on 146 patients with glioblastoma (where 56 patients older than 65 years were included), CCI < 2 did not correlate with outcome. However, only 8/56 elderly patients had a CCI > 2, which made this association uncertain [4]. Conversely, in another small study on 35 patients older than 65 years treated with radio-chemotherapy, age-adjusted CCI correlated with prognosis: patients with CCI < 3 (22/35, 62.8%) had a mOS of 22 months, whereas those with CCI ≥ 3 (13/35, 37.2%) had a mOS of 10 months [5]. However, overall survival of elderly patients with CCI < 3 was longer than what has been reported in previous series; second, the number of patients with CCI ≥ 3 was smaller than usually seen in similar series; third, other clinical and molecular factors (i.e., MGMTp methylation), which might have clarified these issues, were not investigated [5]. In another study on 233 elderly patients, CCI > 3 correlated with worse mPFS and mOS [6]. However, patients had a lower median age than those of our series (62 years vs. 71 years), with similar prevalence of patients with CCI > 3 (about 46% vs. 45.6%) and same median CCI score (3, as in our case). This resembles what we observed in our cohort, where CCI > 3 retained a prognostic importance only in the 65-69-year-old class of age; this was probably due to the impact of comorbidities on outcome in younger rather than older patients who have an independent risk for dismal prognosis and very short survival (Table 6; Supplementary Table S2).
In our study, KPS influenced the choice of treatments (Table 4) and correlated with outcome. Overall, most patients had KPS ≥ 70 at diagnosis. However, patients not undergoing surgery or adjuvant treatments had worse KPS (Table 4). In the multivariable analysis (Table 9), KPS ≥ 70 at baseline favourably impacted the overall survival of the whole cohort (OR 0.487, 0.214-0.908 95% CI). Therefore, similar to other series [7][8][9], KPS was effective in identifying patients with poorer outcomes.
However, other factors had a heavy impact on prognosis. In our series we suggested that surgical resection, as large as possible, represents one of the strongest prognostic factors (Tables 7 and 9). The prognostic role of the extent of surgery among elderly patients with GBM has long been debated, due to concern for a higher risk of surgical-related complications in such a frail population. Evidence of feasibility and safety of extended resections in elderly GBM patients mainly derives from small case series, whereas data from phase III trials are lacking. Preliminary data about the superiority of open craniotomy over stereotactic biopsy in patients aged more than 65 initially came from small retrospective or prospective studies on 30-40 patients [9,10]. Then, a larger retrospective trial on 142 elderly patients with newly diagnosed GBM established the superiority of extended resection over biopsy in terms of overall survival (13.0 months vs. 4.0 months, p < 0.001) [11]. Finally, data from randomised phase III trials NOA-08 and Nordic confirmed that surgical resection is superior to biopsy alone [12,13]. In our study, patients undergoing extended surgery had longer mPFS and mOS (Table 7). Moreover, EOR retained a significant impact on prognosis, being mOS after gross-total resection (GTR) of 32.8 months vs. 9.9 after subtotal resection and 6.0 after biopsy (Table 7). Survival after GTR was remarkably long, which was also explained by the fact that patients undergoing GTR were more frequently treated with combined radio-chemotherapy regimens (Table 4). Additionally, the incidence of clinical complications was not higher in case of GTR, thus proving that GTR was a safe option for elderly patients. As in the multivariable analysis, the EOR retains an independently favourable prognostic role (Table 9), and our study confirms that gross-total resection (GTR), if feasible, is the best option for GBM patients [14][15][16], regardless of molecular status [17].
Adjuvant treatments after surgical resection dramatically impacted the outcome. In our study, we confirmed that combined RT/TMZ followed by TMZ provides the longest survivals in the whole cohort, regardless of MGMTp methylation status (Table 7). However, among patients with MGMTp methylation, survival was longer with RT/TMZ followed by adjuvant TMZ, whereas among those without MGMTp methylation combined radiochemotherapy did not provide a significant benefit (Table 7). These data are consistent with the findings of Perry et al. [18], even if, in our case, the 3-week RT/TMZ regimen was superior to the standard 6-week RT/TMZ only among MGMTp-unmethylated patients, whereas among MGMTp-methylated patients 6-week combined RT/TMZ was still associated with longer overall survival (Table 8).
Lastly, similar to the findings of the Nordic trial [13], in our study, MGMTp-unmethylated patients had a better survival when treated with hypofractioned RT, as compared to upfront CT with TMZ (16.1, 12.9, 6.0, and 4.6 months in the 3-week RT + TMZ, 3-week RT/TMZ + TMZ, hypofractioned RT alone, and upfront TMZ groups, respectively- Table 8), which suggests that the use of combined radio-chemotherapy schedule in this molecular subgroup might be avoided to improve tolerability.
Even if the general prognosis of the whole cohort remains poor (similar to other series), we recognised a subgroup with better outcomes, with remarkably longer mPFS and mOS, characterised by the presence of MGMTp methylation, and after gross-total resection and adjuvant radio-chemotherapy. The implementation of a comprehensive geriatric score would be of primary importance to suggest the best treatment modality to improve survival and avoid treatment-related adverse events among elderly people with frail conditions who may not tolerate aggressive treatments [34].    Abbreviations: BSC, best supportive care; CAD, coronary artery disease; CCI, Charlson Comorbidity Index; CR, complete resection; CT, chemotherapy; DM, diabetes mellitus; DVT, deep venous thrombosis; EOR, extent of resection; pts, patients; KPS, Karnofsky Performance Status; GTR, gross-total resection; met, methylated; MGMT, O(6)-methylguanyl DNA methyltransferase promoter; n.a., not available (i.e., not specified in the paper); OS, overall survival; PFS, progression-free survival; pts, patients; RT, radiotherapy; STR, subtotal resection; TMZ, temozolomide; unmet, unmethylated; wks, weeks.

Conclusions
Elderly GBM patients are generally characterised by limited response to treatment and poor outcomes. However, in our study we identify a subgroup of patients with favourable clinical and molecular features who significantly benefit from large surgical resection and radio-chemotherapy. A comprehensive evaluation of clinical and molecular characteristics of elderly GBM patients is essential to choose the best treatment modality and predict the outcome, in order to increase survival, reduce toxicity, and improve quality of life. Author Contributions: Conceptualization, F.B., A.P., E.P., R.P. and R.R.; methodology, F.B., A.P., E.P., R.P. and R.R.; software, F.B.; validation, F.B., A.P., E.P., R.P., L.B., C.M., A.B., A.M., D.G. and R.R.; formal analysis, F.B. and R.R.; investigation, F.B., A.P., E.P., R.P. and R.R.; resources, F.B. and R.R.; data curation, F.B.; writing-original draft preparation, F.B., A.P. and R.R.; writing-review and editing, F.B., A.P., R.P., C.M., D.G. and R.R.; supervision, R.R.; project administration, R.R. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Ethical review and approval were waived for this study due to the retrospective nature of the study.
Informed Consent Statement: Informed consent to collect and analyse clinical and pathological/molecular data was obtained from all subjects who were alive at the time of start of the study or from relatives in case of death of the subjects, according to ethic regulations for retrospective studies of the local Ethics Committee.