1,420 adult glioblastoma cases met the inclusion criteria. Table 1 provides a summary of basic clinical characteristics of the patient cohort, of IDH- and MGMT promoter status, and of overall survival.

813 (57.3%) patients were male (M) and 607 (42.7%) female (F), the M/F-ratio was 1.3. Median age at diagnosis was 64 years (range: 18–88 years).

IDH-status was wildtype in 1,115/1,420 (78.5%), IDH-mutation was detected in 49 (3.5%), and in 256 (18.0%) IDH-status was unknown (either not analyzed, or result was not available during data collection). MGMT-promoter was methylated in 487 (34.3%), whereas 504 (35.5%) showed no MGMT-promoter-methylation. MGMT-status was unknown in 429 (30.2%).

Most patients presented with focal neurological deficits (599, 42.2%), followed by headache (245, 17.3%), epilepsy (221, 15.6%), personality changes (156, 10.9%), and other presenting symptoms (162, 11.4%; in 37 cases (2.6%), presenting symptoms were unknown). Other presenting symptoms/reasons for patient presentation included: incidental detection of glioblastoma during hospitalization, or as part of the staging of preexisting malignancies; emergency indication of resection for decompression (to reduce intracranial hypertension, e.g., due to an acute intracerebral hemorrhage), and subsequent histological diagnosis.

In Fig. 2A, timespans from the first presenting symptoms to the first diagnostic imaging (MRI) in days are displayed. Depending on the type of symptom, median time to diagnostic imaging ranged from 4 days in patients with epilepsy to 13 days in patients with personality changes.

Clinical time intervals. A (upper figure) Time between first symptoms and scan. Boxplots show the distribution of time intervals (in days) between different types of first symptoms and first diagnostic scan with MRI. Median values vary between symptom categories (shortest interval in epileptic patients, longest interval in patients with personality changes). B (lower figure) Time between scan and surgery. Boxplots as previously described for distribution of time intervals (in days) between first diagnostic scan and first resection. Median values show the shortest interval in patients with personality changes, and the longest interval in epileptic patients (vice versa to A)

Figure 2B shows the time periods from date of first diagnostic imaging to date of first neurosurgical operation. Likewise, timespans differed between presenting symptoms, e.g., 7 days in patients with headache vs. 12 days in case of epileptic seizures.

The initial surgery was either gross total resection (GTR), subtotal resection (STR), or biopsy. The extent of resection (GTR, STR) was rated by the operating neurosurgeon.

GTR was achieved in 579 patients (40.8%), 396 (27.9%) had STR, and 423 (29.8%) underwent a biopsy procedure. In 22 cases (1.5%), the surgical approach is unknown (information not available during data collection). 5-ALA was used in each type of surgery, most commonly in GTR (390, 67.4% of all GTR patients), followed by STR (195, 49.2% of all STR patients) and biopsy (124, 29.3% of all biopsied patients). In 2014, 5-ALA-assisted neurosurgery was applied to 38.3% of all patients, and by 2018 its use had increased to 55.0%.

Most patients started postsurgical treatment approximately one month (median 31 days, 95% confidence interval (CI) 24–44 days) after first neurosurgical procedure. 107 patients (7.5%) could not receive any postsurgical treatment or did not consent to radiotherapy or to chemotherapy. In further 272 patients (19.2%), therapy data were missing due to the following reasons: decease before potential start of post-surgical treatment; patients lost to follow-up; missing data / information on therapy not available during data collection.

Of all 1,420 patients, 1,041 (73.3%) started first line post-surgical treatment. 830 (58.5%) started radiochemotherapy following the SOC protocol according to Stupp et al. [4]: 60 Gray (Gy) in 30 fractions of 2 Gy, combined with temozolomide (TMZ). Other radio-oncological treatment strategies included: the Canadian Cancer Trials Group CCTG CE.6 randomized clinical trial protocol designed for elderly glioblastoma patients (N = 44, 3.1%) [23]; the hypofractionated scheme with 34 Gy from the Nordic trial (N = 71, 5.0%) [24]; as well as other individually adjusted radiotherapy protocols (Fig. 3).

Radio-oncological treatment strategies in relation to patient age. Stacked bar chart showing percentage distribution of radio-oncological therapy regimens in different age groups of Austrian glioblastoma patients. In the younger age groups, relatively high numbers of patients were treated according to standard of care, whereas elderly patients receive more frequently age-adjusted treatment

Of 830 patients (58.5% of the whole patient cohort), who started radiochemotherapy according to the SOC, completion of treatment was documented in 302 patients (21.3%). The remaining 528 patients (37.2%) could not complete therapy due to progression of disease and/or therapy toxicity. In addition, tumor treating fields (TTF) were applied in 43 patients (3.1%). First-line treatment strategies were similar throughout centers across Austria. Disease progression, as defined by the initiation of second-line treatment, was diagnosed in 691 patients (48.7%). Time to progression was 7.8 months in median (95% CI 6.5–8.0 months).

Treatment upon relapse showed great variability. 170 patients (12.0%) underwent a second surgical procedure, 467 (33.0%) received systemic treatment after progression, and 173 (12.2%) were re-irradiated.

Upon disease progression, 424 (29.9%) received anti-angiogenic therapy.

Median overall survival (OS) of all glioblastoma patients in this study cohort was 11.6 months (95% CI 10.7–12.3), and 10.9 months (95% CI 10.3–11.8) in the cohort with proven IDH-wildtype (1,115/1,420). Median OS in the patient group ≤ 65 years receiving postoperative SOC therapy according to Stupp et al. [4] was 16.1 months (95% CI 15.5–17.2). In patients older than > 65 years receiving any kind of postoperative therapy, median OS was 11.2 months (95% CI 10.4–12.2). Figure 4 shows survival relative to specific prognostic factors, including sex, patient age, extent of resection, IDH-status, MGMT-promoter-methylation, ECOG score, and TTF therapy. A peculiar finding is the relatively high fraction of patients in whom the MGMT status is unknown. In these cases, explanatory information for the lack of MGMT testing is most often missing, e.g., if testing was not done for particular clinical reasons; or MGMT testing was not available in all centers at the beginning of the registry; or if the amount of biopsy tissue available was not sufficient for testing. Of note, survival in the small cohort of glioblastoma patients with TTF-therapy (N = 43) showed a significantly better outcome than patients without TTF-therapy.

Results of univariate survival analyses. Kaplan–Meier plots showing overall survival A as well as analyses of specific prognostic factors (B-H). Kaplan–Meier plots are accompanied by bar charts showing numbers of patients in different categories. B Sex: there is no sex difference with regard to outcome. C Age: there is a clear indirect correlation: young patients show a significantly more favorable outcome as compared to elderly patients. D Extent of resection: there is a direct correlation: patients with gross-total/subtotal resection survive significantly better than patients with biopsy only. E IDH-status: significantly better survival in IDH-mutated cases. F MGMT-promoter-methylation: significantly better survival in cases with MGMT-promoter methylation. G ECOG score: there is a clear indirect correlation: patients with a low postoperative ECOG score survive significantly better as compared to patients with a high ECOG score. H Tumor treating fields (TTF) therapy: the small group of patients who received TTF-therapy show a significantly better outcome

In Fig. 5, multivariate survival analyses using Cox’s regression model are displayed in a forest plot. The total number of analyzed patients is 1,398/1,420 patients, because in 22 patients the extent of resection was unknown, and this small group was excluded to avoid any distortion of results. Analyses are split into three intervals: Short OS (less than 3 months), Moderate OS (3 to 24 month) and Long OS (more than 24 months), in order to learn the relative impact of various patient parameters on outcome over time. The following parameters were analyzed: sex, age, extent of resection, MGMT methylation status, IDH status, postoperative ECOG performance status, re-resection, and TTF.

Results of multivariate survival analyses. Forest plots showing results of multivariate survival analyses for specific prognostic factors in three different groups of overall survival (OS): Short OS (< 3 months), Moderate OS (3 to 24 months) and Long OS (> 24 months). Statistically significant values are indicated with one (slight significance) or three asterisks (high significance). In the Short OS group, the strongest independent predictor for survival is the ECOG score. In the Moderate OS group, factors with high independent impact on survival comprise: patient age; extent of resection; MGMT promoter methylation status; ECOG score; reoperation; and use of TTF. In the Long OS group, the single strong independent predictor for survival is patient age

No statistically significant differences in OS are detected between female and male patients.

Comparison of age (split into two groups, < 65 and ≥ 65 years) shows a statistically significant risk for elderly patients, especially in the moderate and long OS intervals. Further, statistically significant results of hazard ratio in different OS intervals include: beneficial effects of gross total resection (GTR) and subtotal resection (STR) in the short and moderate OS intervals compared to biopsies; unfavorable outcome of patients with unmethylated MGMT promoter status (compared to methylated) in the moderate OS interval; unfavorable patient outcome associated with worsening postoperative ECOG performance score in the short and moderate OS intervals; favorable outcome effects in patients with re-resection compared to those with only one operation in the short and moderate OS intervals; favorable outcome effect in patients with use of TTF in the moderate OS interval (for further details see Fig. 5).

A slight statistically significant difference was found regarding IDH status within the group with moderate overall survival, depicting a higher risk (hazard ratio 1.71, 1.01–2.87) for patients with IDH wildtype compared to IDH mutation.

The maximum follow-up observation period in our study was 6 years, and only the subcohort of 264/1,420 patients included in the first year of data registration (in 2014) can be considered as long-term (≥ 5 years) follow-up group. Long-term survival (≥ 5 years) was observed in 13/264 patients (4.9%). The majority of patients in this small group of long-term survivors belonged to the younger age group (< 65 years: 11/13), usually had gross total tumor resection (10/13), a favorable postoperative ECOG score (ECOG 0–2 in 12/13 cases, in 1/13 unknown), a total tumor resection (10/13), and postoperative standard of care therapy (13/13); 5/13 were female, and 8/13 were male. MGMT promoter was methylated in 7/13, unmethylated in 1/13, and in 5/13, the methylation status was unknown. IDH was wildtype in 8/13 patients, and in 5/13 patients, the IDH status was unknown.